# REST API Reference

 Input AboutV3 Output AboutV3

## GET /3/Cloud

Determine the status of the nodes in the H2O cloud.

 Input CloudV3 Output CloudV3

Determine the status of the nodes in the H2O cloud.

 Input CloudV3 Output CloudV3

## GET /3/ComputeGram

Get weighted gram matrix

 Input GramV3 Output GramV3

## POST /3/CreateFrame

Create a synthetic H2O Frame with random data. You can specify the number of rows/columns, as well as column types: integer, real, boolean, time, string, categorical. The frame may also have a dedicated “response” column, and some of the entries in the dataset may be created as missing.

 Input CreateFrameV3 Output JobV3

## DELETE /3/DKV

Remove all keys from the H2O distributed K/V store.

 Input RemoveAllV3 Output RemoveAllV3

## DELETE /3/DKV/{key}

Remove an arbitrary key from the H2O distributed K/V store.

 Input RemoveV3 Output RemoveV3

## POST /3/DataInfoFrame

Test only

 Input DataInfoFrameV3 Output DataInfoFrameV3

 Input DownloadDataV3 Output DownloadDataV3

 Input DownloadDataV3 Output DownloadDataV3

## GET /3/Find

Find a value within a Frame.

 Input FindV3 Output FindV3

## GET /3/Frames

Return all Frames in the H2O distributed K/V store.

 Input FramesV3 Output FramesV3

## DELETE /3/Frames

Delete all Frames from the H2O distributed K/V store.

 Input FramesV3 Output FramesV3

## GET /3/Frames/{frame_id}

Return the specified Frame.

 Input FramesV3 Output FramesV3

## DELETE /3/Frames/{frame_id}

Delete the specified Frame from the H2O distributed K/V store.

 Input FramesV3 Output FramesV3

## GET /3/Frames/{frame_id}/columns

Return all the columns from a Frame.

 Input FramesV3 Output FramesV3

## GET /3/Frames/{frame_id}/columns/{column}

Return the specified column from a Frame.

 Input FramesV3 Output FramesV3

## GET /3/Frames/{frame_id}/columns/{column}/domain

Return the domains for the specified categorical column (“null” if the column is not a categorical).

 Input FramesV3 Output FramesV3

## GET /3/Frames/{frame_id}/columns/{column}/summary

Return the summary metrics for a column, e.g. min, max, mean, sigma, percentiles, etc.

 Input FramesV3 Output FramesV3

## POST /3/Frames/{frame_id}/export

Export a Frame to the given path with optional overwrite.

 Input FramesV3 Output FramesV3

## GET /3/Frames/{frame_id}/export/{path}/overwrite/{force}

[DEPRECATED] Export a Frame to the given path with optional overwrite.

 Input FramesV3 Output FramesV3

## GET /3/Frames/{frame_id}/summary

Return a Frame, including the histograms, after forcing computation of rollups.

 Input FramesV3 Output FramesV3

## POST /3/GarbageCollect

Explicitly call System.gc().

 Input GarbageCollectV3 Output GarbageCollectV3

## GET /3/GetGLMRegPath

Get full regularization path

 Input GLMRegularizationPathV3 Output GLMRegularizationPathV3

## GET /3/ImportFiles

[DEPRECATED] Import raw data files into a single-column H2O Frame.

 Input ImportFilesV3 Output ImportFilesV3

## POST /3/ImportFiles

Import raw data files into a single-column H2O Frame.

 Input ImportFilesV3 Output ImportFilesV3

## GET /3/InitID

Issue a new session ID.

 Input InitIDV3 Output InitIDV3

## DELETE /3/InitID

End a session.

 Input InitIDV3 Output InitIDV3

## POST /3/Interaction

Create interactions between categorical columns.

 Input InteractionV3 Output JobV3

## GET /3/JStack

Report stack traces for all threads on all nodes.

 Input JStackV3 Output JStackV3

## GET /3/Jobs

Get a list of all the H2O Jobs (long-running actions).

 Input JobsV3 Output JobsV3

## GET /3/Jobs/{job_id}

Get the status of the given H2O Job (long-running action).

 Input JobsV3 Output JobsV3

## POST /3/Jobs/{job_id}/cancel

Cancel a running job.

 Input JobsV3 Output JobsV3

## GET /3/KillMinus3

Kill minus 3 on this node

 Input KillMinus3V3 Output KillMinus3V3

## POST /3/LogAndEcho

Save a message to the H2O logfile.

 Input LogAndEchoV3 Output LogAndEchoV3

## GET /3/Logs/nodes/{nodeidx}/files/{name}

Get named log file for a node.

 Input LogsV3 Output LogsV3

## POST /3/MakeGLMModel

Make a new GLM model based on existing one

 Input MakeGLMModelV3 Output GLMModelV3

Return the list of (almost) all REST API endpoints.

 Input MetadataV3 Output MetadataV3

Return the REST API endpoint metadata, including documentation, for the endpoint specified by path or index.

 Input MetadataV3 Output MetadataV3

Return the REST API schema metadata for specified schema class.

 Input MetadataV3 Output MetadataV3

Return list of all REST API schemas.

 Input MetadataV3 Output MetadataV3

Return the REST API schema metadata for specified schema.

 Input MetadataV3 Output MetadataV3

## POST /3/MissingInserter

Insert missing values.

 Input MissingInserterV3 Output JobV3

## GET /3/ModelBuilders

Return the Model Builder metadata for all available algorithms.

 Input ModelBuildersV3 Output ModelBuildersV3

## POST /3/ModelBuilders/deeplearning

Train a DeepLearning model.

 Input DeepLearningV3 Output DeepLearningV3

## POST /3/ModelBuilders/deeplearning/parameters

Validate a set of DeepLearning model builder parameters.

 Input DeepLearningV3 Output DeepLearningV3

## POST /3/ModelBuilders/deepwater

Train a DeepWater model.

 Input DeepWaterV3 Output DeepWaterV3

## POST /3/ModelBuilders/deepwater/parameters

Validate a set of DeepWater model builder parameters.

 Input DeepWaterV3 Output DeepWaterV3

## POST /3/ModelBuilders/drf

Train a DRF model.

 Input DRFV3 Output DRFV3

## POST /3/ModelBuilders/drf/parameters

Validate a set of DRF model builder parameters.

 Input DRFV3 Output DRFV3

## POST /3/ModelBuilders/gbm

Train a GBM model.

 Input GBMV3 Output GBMV3

## POST /3/ModelBuilders/gbm/parameters

Validate a set of GBM model builder parameters.

 Input GBMV3 Output GBMV3

## POST /3/ModelBuilders/glm

Train a GLM model.

 Input GLMV3 Output GLMV3

## POST /3/ModelBuilders/glm/parameters

Validate a set of GLM model builder parameters.

 Input GLMV3 Output GLMV3

## POST /3/ModelBuilders/glrm

Train a GLRM model.

 Input GLRMV3 Output GLRMV3

## POST /3/ModelBuilders/glrm/parameters

Validate a set of GLRM model builder parameters.

 Input GLRMV3 Output GLRMV3

## POST /3/ModelBuilders/kmeans

Train a KMeans model.

 Input KMeansV3 Output KMeansV3

## POST /3/ModelBuilders/kmeans/parameters

Validate a set of KMeans model builder parameters.

 Input KMeansV3 Output KMeansV3

## POST /3/ModelBuilders/naivebayes

Train a NaiveBayes model.

 Input NaiveBayesV3 Output NaiveBayesV3

## POST /3/ModelBuilders/naivebayes/parameters

Validate a set of NaiveBayes model builder parameters.

 Input NaiveBayesV3 Output NaiveBayesV3

## POST /3/ModelBuilders/pca

Train a PCA model.

 Input PCAV3 Output PCAV3

## POST /3/ModelBuilders/pca/parameters

Validate a set of PCA model builder parameters.

 Input PCAV3 Output PCAV3

## POST /3/ModelBuilders/word2vec

Train a Word2Vec model.

 Input Word2VecV3 Output Word2VecV3

## POST /3/ModelBuilders/word2vec/parameters

Validate a set of Word2Vec model builder parameters.

 Input Word2VecV3 Output Word2VecV3

## GET /3/ModelBuilders/{algo}

Return the Model Builder metadata for the specified algorithm.

 Input ModelBuildersV3 Output ModelBuildersV3

## POST /3/ModelBuilders/{algo}/model_id

Return a new unique model_id for the specified algorithm.

 Input ModelBuildersV3 Output ModelIdV3

## GET /3/ModelMetrics

Return all the saved scoring metrics.

 Input ModelMetricsListSchemaV3 Output ModelMetricsListSchemaV3

## GET /3/ModelMetrics/frames/{frame}

Return the saved scoring metrics for the specified Frame.

 Input ModelMetricsListSchemaV3 Output ModelMetricsListSchemaV3

## GET /3/ModelMetrics/frames/{frame}/models/{model}

Return the saved scoring metrics for the specified Model and Frame.

 Input ModelMetricsListSchemaV3 Output ModelMetricsListSchemaV3

## DELETE /3/ModelMetrics/frames/{frame}/models/{model}

Return the saved scoring metrics for the specified Model and Frame.

 Input ModelMetricsListSchemaV3 Output ModelMetricsListSchemaV3

## GET /3/ModelMetrics/models/{model}

Return the saved scoring metrics for the specified Model.

 Input ModelMetricsListSchemaV3 Output ModelMetricsListSchemaV3

## GET /3/ModelMetrics/models/{model}/frames/{frame}

Return the saved scoring metrics for the specified Model and Frame.

 Input ModelMetricsListSchemaV3 Output ModelMetricsListSchemaV3

## DELETE /3/ModelMetrics/models/{model}/frames/{frame}

Return the saved scoring metrics for the specified Model and Frame.

 Input ModelMetricsListSchemaV3 Output ModelMetricsListSchemaV3

## POST /3/ModelMetrics/models/{model}/frames/{frame}

Return the scoring metrics for the specified Frame with the specified Model. If the Frame has already been scored with the Model then cached results will be returned; otherwise predictions for all rows in the Frame will be generated and the metrics will be returned.

 Input ModelMetricsListSchemaV3 Output ModelMetricsListSchemaV3

## POST /3/ModelMetrics/predictions_frame/{predictions_frame}/actuals_frame/{actuals_frame}

Create a ModelMetrics object from the predicted and actual values, and a domain for classification problems or a distribution family for regression problems.

 Input ModelMetricsMakerSchemaV3 Output ModelMetricsMakerSchemaV3

## GET /3/Models

Return all Models from the H2O distributed K/V store.

 Input ModelsV3 Output ModelsV3

## DELETE /3/Models

Delete all Models from the H2O distributed K/V store.

 Input ModelsV3 Output ModelsV3

## GET /3/Models.java/{model_id}

[DEPRECATED] Return the stream containing model implementation in Java code.

 Input ModelsV3 Output StreamingSchema

## GET /3/Models.java/{model_id}/preview

Return potentially abridged model suitable for viewing in a browser (currently only used for java model code).

 Input ModelsV3 Output StreamingSchema

## GET /3/Models/{model_id}

Return the specified Model from the H2O distributed K/V store, optionally with the list of compatible Frames.

 Input ModelsV3 Output ModelsV3

## DELETE /3/Models/{model_id}

Delete the specified Model from the H2O distributed K/V store.

 Input ModelsV3 Output ModelsV3

## GET /3/Models/{model_id}/mojo

Return the model in the MOJO format. This format can then be interpreted by gen_model.jar in order to perform prediction / scoring. Currently works for GBM and DRF algos only.

 Input ModelsV3 Output StreamingSchema

## GET /3/NetworkTest

Run a network test to measure the performance of the cluster interconnect.

 Input NetworkTestV3 Output NetworkTestV3

## GET /3/NodePersistentStorage/categories/{category}/exists

Return true or false.

 Input NodePersistentStorageV3 Output NodePersistentStorageV3

## GET /3/NodePersistentStorage/categories/{category}/names/{name}/exists

Return true or false.

 Input NodePersistentStorageV3 Output NodePersistentStorageV3

## GET /3/NodePersistentStorage/configured

Return true or false.

 Input NodePersistentStorageV3 Output NodePersistentStorageV3

## POST /3/NodePersistentStorage/{category}

Store a value.

 Input NodePersistentStorageV3 Output NodePersistentStorageV3

## GET /3/NodePersistentStorage/{category}

Return all keys stored for a given category.

 Input NodePersistentStorageV3 Output NodePersistentStorageV3

## POST /3/NodePersistentStorage/{category}/{name}

Store a named value.

 Input NodePersistentStorageV3 Output NodePersistentStorageV3

## GET /3/NodePersistentStorage/{category}/{name}

Return value for a given name.

 Input NodePersistentStorageV3 Output NodePersistentStorageV3

## DELETE /3/NodePersistentStorage/{category}/{name}

Delete a key.

 Input NodePersistentStorageV3 Output NodePersistentStorageV3

## POST /3/Parse

Parse a raw byte-oriented Frame into a useful columnar data Frame.

 Input ParseV3 Output ParseV3

## POST /3/ParseSVMLight

Parse a raw byte-oriented Frame into a useful columnar data Frame.

 Input ParseSVMLightV3 Output JobV3

## POST /3/ParseSetup

Guess the parameters for parsing raw byte-oriented data into an H2O Frame.

 Input ParseSetupV3 Output ParseSetupV3

## POST /3/PartialDependence/

Create data for partial dependence plot(s) for the specified model and frame.

 Input PartialDependenceV3 Output JobV3

## GET /3/PartialDependence/{name}

Fetch partial dependence data.

 Input PartialDependenceKeyV3 Output PartialDependenceV3

## POST /3/Predictions/models/{model}/frames/{frame}

Score (generate predictions) for the specified Frame with the specified Model. Both the Frame of predictions and the metrics will be returned.

 Input ModelMetricsListSchemaV3 Output ModelMetricsListSchemaV3

## GET /3/Profiler

Report real-time profiling information for all nodes (sorted, aggregated stack traces).

 Input ProfilerV3 Output ProfilerV3

## POST /3/Shutdown

Shut down the cluster.

 Input ShutdownV3 Output ShutdownV3

## POST /3/SplitFrame

Split an H2O Frame.

 Input SplitFrameV3 Output SplitFrameV3

## GET /3/Timeline

Debugging tool that provides information on current communication between nodes.

 Input TimelineV3 Output TimelineV3

 Input TypeaheadV3 Output TypeaheadV3

## POST /3/UnlockKeys

Unlock all keys in the H2O distributed K/V store, to attempt to recover from a crash.

 Input UnlockKeysV3 Output UnlockKeysV3

## GET /3/WaterMeterCpuTicks/{nodeidx}

Return a CPU usage snapshot of all cores of all nodes in the H2O cluster.

 Input WaterMeterCpuTicksV3 Output WaterMeterCpuTicksV3

## GET /3/WaterMeterIo

Return IO usage snapshot of all nodes in the H2O cluster.

 Input WaterMeterIoV3 Output WaterMeterIoV3

## GET /3/WaterMeterIo/{nodeidx}

Return IO usage snapshot of all nodes in the H2O cluster.

 Input WaterMeterIoV3 Output WaterMeterIoV3

## GET /3/Word2VecSynonyms

Find synonyms using a word2vec model

 Input Word2VecSynonymsV3 Output Word2VecSynonymsV3

## GET /3/Word2VecTransform

Transform words to vectors using a word2vec model

 Input Word2VecTransformV3 Output Word2VecTransformV3

## POST /4/Predictions/models/{model}/frames/{frame}

Score (generate predictions) for the specified Frame with the specified Model. Both the Frame of predictions and the metrics will be returned.

 Input ModelMetricsListSchemaV3 Output JobV3

## GET /4/endpoints

Returns the list of all REST API (v4) endpoints.

 Input ListRequestV4 Output EndpointsListV4

## GET /4/modelsinfo

Return basic information about all models available to train.

 Input ListRequestV4 Output ModelsInfoV4

## POST /4/sessions

Start a new Rapids session, and return the session id.

 Input InputSchemaV4 Output SessionIdV4

## DELETE /4/sessions/{session_key}

Close the Rapids session.

 Input InitIDV3 Output InitIDV3

## POST /99/Assembly

Fit an assembly to an input frame

 Input AssemblyV99 Output AssemblyV99

## GET /99/Assembly.java/{assembly_id}/{pojo_name}

Generate a Java POJO from the Assembly

 Input AssemblyV99 Output AssemblyV99

## POST /99/DCTTransformer

Row-by-row discrete cosine transforms in 1D, 2D and 3D.

 Input DCTTransformerV3 Output JobV3

## POST /99/Grid/aggregator

Run grid search for Aggregator model.

 Input AggregatorV99 Output AggregatorV99

## POST /99/Grid/deeplearning

Run grid search for DeepLearning model.

 Input DeepLearningV3 Output DeepLearningV3

## POST /99/Grid/deepwater

Run grid search for DeepWater model.

 Input DeepWaterV3 Output DeepWaterV3

## POST /99/Grid/drf

Run grid search for DRF model.

 Input DRFV3 Output DRFV3

## POST /99/Grid/gbm

Run grid search for GBM model.

 Input GBMV3 Output GBMV3

## POST /99/Grid/glm

Run grid search for GLM model.

 Input GLMV3 Output GLMV3

## POST /99/Grid/glrm

Run grid search for GLRM model.

 Input GLRMV3 Output GLRMV3

## POST /99/Grid/kmeans

Run grid search for KMeans model.

 Input KMeansV3 Output KMeansV3

## POST /99/Grid/naivebayes

Run grid search for NaiveBayes model.

 Input NaiveBayesV3 Output NaiveBayesV3

## POST /99/Grid/pca

Run grid search for PCA model.

 Input PCAV3 Output PCAV3

## POST /99/Grid/svd

Run grid search for SVD model.

 Input SVDV99 Output SVDV99

## POST /99/Grid/word2vec

Run grid search for Word2Vec model.

 Input Word2VecV3 Output Word2VecV3

## GET /99/Grids

Return all grids from H2O distributed K/V store.

 Input GridsV99 Output GridsV99

## GET /99/Grids/{grid_id}

Return the specified grid search result.

 Input GridSchemaV99 Output GridSchemaV99

## POST /99/ImportSQLTable

Import SQL table into an H2O Frame.

 Input ImportSQLTableV99 Output JobV3

## POST /99/ModelBuilders/aggregator

Train a Aggregator model.

 Input AggregatorV99 Output AggregatorV99

## POST /99/ModelBuilders/aggregator/parameters

Validate a set of Aggregator model builder parameters.

 Input AggregatorV99 Output AggregatorV99

## POST /99/ModelBuilders/svd

Train a SVD model.

 Input SVDV99 Output SVDV99

## POST /99/ModelBuilders/svd/parameters

Validate a set of SVD model builder parameters.

 Input SVDV99 Output SVDV99

## POST /99/Models.bin/{model_id}

Import given binary model into H2O.

 Input ModelImportV3 Output ModelsV3

## GET /99/Models.bin/{model_id}

Export given model.

 Input ModelExportV3 Output ModelExportV3

## GET /99/Models.mojo/{model_id}

Export given model as Mojo.

 Input ModelExportV3 Output ModelExportV3

## POST /99/Rapids

Execute an Rapids AstRoot.

 Input RapidsSchemaV3 Output RapidsSchemaV3

## GET /99/Rapids/help

Produce help for Rapids AstRoot language.

 Input SchemaV3 Output RapidsHelpV3

## GET /99/Sample

Example of an experimental endpoint. Call via /EXPERIMENTAL/Sample. Experimental endpoints can change at any moment.

 Input CloudV3 Output CloudV3

## POST /99/Tabulate

Tabulate one column vs another.

 Input TabulateV3 Output TabulateV3

# REST API Schema Reference

 namestring Property name Out valuestring Property value Out

 _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In entriesIced[] List of properties about this running H2O instance Out

## AggregatorModelOutputV99

 output_frameKey Aggregated Frame of Exemplars In namesstring[] Column names Out domainsstring[][] Domains for categorical columns Out cross_validation_modelsKey[] Cross-validation models (model ids) Out cross_validation_predictionsKey[] Cross-validation predictions, one per cv model (deprecated, use cross_validation_holdout_predictions_frame_id instead) Out cross_validation_holdout_predictions_frame_idKey Cross-validation holdout predictions (full out-of-sample predictions on training data) Out cross_validation_fold_assignment_frame_idKey Cross-validation fold assignment (each row is assigned to one holdout fold) Out model_categoryenum Category of the model (e.g., Binomial) Out model_summaryTwoDimTable Model summary Out scoring_historyTwoDimTable Scoring history Out training_metricsModelMetrics Training data model metrics Out validation_metricsModelMetrics Validation data model metrics Out cross_validation_metricsModelMetrics Cross-validation model metrics Out cross_validation_metrics_summaryTwoDimTable Cross-validation model metrics summary Out statusstring Job status Out start_timelong Start time in milliseconds Out end_timelong End time in milliseconds Out run_timelong Runtime in milliseconds Out helpMap Help information for output fields Out

## AggregatorModelV99

 model_idKey Model key In/Out parametersAggregatorParameters The build parameters for the model (e.g. K for KMeans). Out outputAggregatorOutput The build output for the model (e.g. the cluster centers for KMeans). Out compatible_framesstring[] Compatible frames, if requested Out checksumlong Checksum for all the things that go into building the Model. Out algostring The algo name for this Model. Out algo_full_namestring The pretty algo name for this Model (e.g., Generalized Linear Model, rather than GLM). Out response_column_namestring The response column name for this Model (if applicable). Is null otherwise. Out data_frameKey The Model’s training frame key Out timestamplong Timestamp for when this model was completed Out

## AggregatorParametersV99

 radius_scaledouble Radius scaling In transformenum Transformation of training data In pca_methodenum Method for computing PCA (Caution: Power and GLRM are currently experimental and unstable) In distributionenum Distribution function In tweedie_powerdouble Tweedie power for Tweedie regression, must be between 1 and 2. In quantile_alphadouble Desired quantile for Quantile regression, must be between 0 and 1. In huber_alphadouble Desired quantile for Huber/M-regression (threshold between quadratic and linear loss, must be between 0 and 1). In kint Rank of matrix approximation In/Out max_iterationsint Maximum number of iterations for PCA In/Out seedlong RNG seed for initialization In/Out use_all_factor_levelsboolean Whether first factor level is included in each categorical expansion In/Out model_idKey Destination id for this model; auto-generated if not specified. In/Out training_frameKey Id of the training data frame (Not required, to allow initial validation of model parameters). In/Out validation_frameKey Id of the validation data frame. In/Out nfoldsint Number of folds for N-fold cross-validation (0 to disable or >= 2). In/Out keep_cross_validation_predictionsboolean Whether to keep the predictions of the cross-validation models. In/Out keep_cross_validation_fold_assignmentboolean Whether to keep the cross-validation fold assignment. In/Out parallelize_cross_validationboolean Allow parallel training of cross-validation models In/Out response_columnVecSpecifier Response variable column. In/Out weights_columnVecSpecifier Column with observation weights. Giving some observation a weight of zero is equivalent to excluding it from the dataset; giving an observation a relative weight of 2 is equivalent to repeating that row twice. Negative weights are not allowed. In/Out offset_columnVecSpecifier Offset column. This will be added to the combination of columns before applying the link function. In/Out fold_columnVecSpecifier Column with cross-validation fold index assignment per observation. In/Out fold_assignmentenum Cross-validation fold assignment scheme, if fold_column is not specified. The ‘Stratified’ option will stratify the folds based on the response variable, for classification problems. In/Out categorical_encodingenum Encoding scheme for categorical features In/Out ignored_columnsstring[] Names of columns to ignore for training. In/Out ignore_const_colsboolean Ignore constant columns. In/Out score_each_iterationboolean Whether to score during each iteration of model training. In/Out checkpointKey Model checkpoint to resume training with. In/Out stopping_roundsint Early stopping based on convergence of stopping_metric. Stop if simple moving average of length k of the stopping_metric does not improve for k:=stopping_rounds scoring events (0 to disable) In/Out max_runtime_secsdouble Maximum allowed runtime in seconds for model training. Use 0 to disable. In/Out stopping_metricenum Metric to use for early stopping (AUTO: logloss for classification, deviance for regression) In/Out stopping_tolerancedouble Relative tolerance for metric-based stopping criterion (stop if relative improvement is not at least this much) In/Out

## AggregatorV99

 parametersAggregatorParameters Model builder parameters. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In algostring The algo name for this ModelBuilder. Out algo_full_namestring The pretty algo name for this ModelBuilder (e.g., Generalized Linear Model, rather than GLM). Out can_buildenum[] Model categories this ModelBuilder can build. Out supervisedboolean Indicator whether the model is supervised or not. Out visibilityenum Should the builder always be visible, be marked as beta, or only visible if the user starts up with the experimental flag? Out jobJob Job Key Out messagesValidationMessage[] Parameter validation messages Out error_countint Count of parameter validation errors Out __http_statusint HTTP status to return for this build. Out

## AssemblyKeyV3

 namestring Name (string representation) for this Key. In/Out typestring Name (string representation) for the type of Keyed this Key points to. In/Out URLstring URL for the resource that this Key points to, if one exists. In/Out

## AssemblyV99

 stepsstring[] A list of steps describing the assembly line. In frameKey Input Frame for the assembly. In pojo_namestring The name of the file and generated class In assembly_idstring The key of the Assembly object to retrieve from the DKV. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In resultKey Output of the assembly line. Out assemblyKey A Key to the fit Assembly data structure Out

## CartesianSearchCriteriaV99

 strategyenum Hyperparameter space search strategy. In/Out

## CloudV3

 skip_ticksboolean skip_ticks In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In versionstring version Out branch_namestring branch_name Out build_numberstring build_number Out build_agestring build_age Out build_too_oldboolean build_too_old Out node_idxint Node index number cloud status is collected from (zero-based) Out cloud_namestring cloud_name Out cloud_sizeint cloud_size Out cloud_uptime_millislong cloud_uptime_millis Out cloud_healthyboolean cloud_healthy Out bad_nodesint Nodes reporting unhealthy Out consensusboolean Cloud voting is stable Out lockedboolean Cloud is accepting new members or not Out is_clientboolean Cloud is in client mode. Out nodesIced[] nodes Out

## ClusteringModelBuilderSchema

 parametersParameters Model builder parameters. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In algostring The algo name for this ModelBuilder. Out algo_full_namestring The pretty algo name for this ModelBuilder (e.g., Generalized Linear Model, rather than GLM). Out can_buildenum[] Model categories this ModelBuilder can build. Out supervisedboolean Indicator whether the model is supervised or not. Out visibilityenum Should the builder always be visible, be marked as beta, or only visible if the user starts up with the experimental flag? Out jobJob Job Key Out messagesValidationMessage[] Parameter validation messages Out error_countint Count of parameter validation errors Out __http_statusint HTTP status to return for this build. Out

 distributionenum Distribution function In tweedie_powerdouble Tweedie power for Tweedie regression, must be between 1 and 2. In quantile_alphadouble Desired quantile for Quantile regression, must be between 0 and 1. In huber_alphadouble Desired quantile for Huber/M-regression (threshold between quadratic and linear loss, must be between 0 and 1). In kint The max. number of clusters. If estimate_k is disabled, the model will find k centroids, otherwise it will find up to k centroids. In/Out model_idKey Destination id for this model; auto-generated if not specified. In/Out training_frameKey Id of the training data frame (Not required, to allow initial validation of model parameters). In/Out validation_frameKey Id of the validation data frame. In/Out nfoldsint Number of folds for N-fold cross-validation (0 to disable or >= 2). In/Out keep_cross_validation_predictionsboolean Whether to keep the predictions of the cross-validation models. In/Out keep_cross_validation_fold_assignmentboolean Whether to keep the cross-validation fold assignment. In/Out parallelize_cross_validationboolean Allow parallel training of cross-validation models In/Out response_columnVecSpecifier Response variable column. In/Out weights_columnVecSpecifier Column with observation weights. Giving some observation a weight of zero is equivalent to excluding it from the dataset; giving an observation a relative weight of 2 is equivalent to repeating that row twice. Negative weights are not allowed. In/Out offset_columnVecSpecifier Offset column. This will be added to the combination of columns before applying the link function. In/Out fold_columnVecSpecifier Column with cross-validation fold index assignment per observation. In/Out fold_assignmentenum Cross-validation fold assignment scheme, if fold_column is not specified. The ‘Stratified’ option will stratify the folds based on the response variable, for classification problems. In/Out categorical_encodingenum Encoding scheme for categorical features In/Out ignored_columnsstring[] Names of columns to ignore for training. In/Out ignore_const_colsboolean Ignore constant columns. In/Out score_each_iterationboolean Whether to score during each iteration of model training. In/Out checkpointKey Model checkpoint to resume training with. In/Out stopping_roundsint Early stopping based on convergence of stopping_metric. Stop if simple moving average of length k of the stopping_metric does not improve for k:=stopping_rounds scoring events (0 to disable) In/Out max_runtime_secsdouble Maximum allowed runtime in seconds for model training. Use 0 to disable. In/Out stopping_metricenum Metric to use for early stopping (AUTO: logloss for classification, deviance for regression) In/Out stopping_tolerancedouble Relative tolerance for metric-based stopping criterion (stop if relative improvement is not at least this much) In/Out

## ColSpecifierV3

 column_namestring Name of the column In/Out is_member_of_framesstring[] List of fields which specify columns that must contain this column In/Out

## ColV3

 labelstring label Out missing_countlong missing Out zero_countlong zeros Out positive_infinity_countlong positive infinities Out negative_infinity_countlong negative infinities Out minsdouble[] mins Out maxsdouble[] maxs Out meandouble mean Out sigmadouble sigma Out typestring datatype: {enum, string, int, real, time, uuid} Out domainstring[] domain; not-null for categorical columns only Out domain_cardinalityint cardinality of this column’s domain; not-null for categorical columns only Out datadouble[] data Out string_datastring[] string data Out precisionbyte decimal precision, -1 for all digits Out histogram_binslong[] Histogram bins; null if not computed Out histogram_basedouble Start of histogram bin zero Out histogram_stridedouble Stride per bin Out percentilesdouble[] Percentile values, matching the default percentiles Out

## ColumnSpecsBase

 namestring Column Name Out typestring Column Type Out formatstring Column Format (printf) Out descriptionstring Column Description Out

## ConfusionMatrixV3

 tableTwoDimTable Annotated confusion matrix Out

## CoxPHModelOutputV3

 namesstring[] Column names Out domainsstring[][] Domains for categorical columns Out cross_validation_modelsKey[] Cross-validation models (model ids) Out cross_validation_predictionsKey[] Cross-validation predictions, one per cv model (deprecated, use cross_validation_holdout_predictions_frame_id instead) Out cross_validation_holdout_predictions_frame_idKey Cross-validation holdout predictions (full out-of-sample predictions on training data) Out cross_validation_fold_assignment_frame_idKey Cross-validation fold assignment (each row is assigned to one holdout fold) Out model_categoryenum Category of the model (e.g., Binomial) Out model_summaryTwoDimTable Model summary Out scoring_historyTwoDimTable Scoring history Out training_metricsModelMetrics Training data model metrics Out validation_metricsModelMetrics Validation data model metrics Out cross_validation_metricsModelMetrics Cross-validation model metrics Out cross_validation_metrics_summaryTwoDimTable Cross-validation model metrics summary Out statusstring Job status Out start_timelong Start time in milliseconds Out end_timelong End time in milliseconds Out run_timelong Runtime in milliseconds Out helpMap Help information for output fields Out

## CoxPHModelV3

 model_idKey Model key In/Out parametersCoxPHParameters The build parameters for the model (e.g. K for KMeans). Out outputCoxPHOutput The build output for the model (e.g. the cluster centers for KMeans). Out compatible_framesstring[] Compatible frames, if requested Out checksumlong Checksum for all the things that go into building the Model. Out algostring The algo name for this Model. Out algo_full_namestring The pretty algo name for this Model (e.g., Generalized Linear Model, rather than GLM). Out response_column_namestring The response column name for this Model (if applicable). Is null otherwise. Out data_frameKey The Model’s training frame key Out timestamplong Timestamp for when this model was completed Out

## CoxPHParametersV3

 distributionenum Distribution function In tweedie_powerdouble Tweedie power for Tweedie regression, must be between 1 and 2. In quantile_alphadouble Desired quantile for Quantile regression, must be between 0 and 1. In huber_alphadouble Desired quantile for Huber/M-regression (threshold between quadratic and linear loss, must be between 0 and 1). In model_idKey Destination id for this model; auto-generated if not specified. In/Out training_frameKey Id of the training data frame (Not required, to allow initial validation of model parameters). In/Out validation_frameKey Id of the validation data frame. In/Out nfoldsint Number of folds for N-fold cross-validation (0 to disable or >= 2). In/Out keep_cross_validation_predictionsboolean Whether to keep the predictions of the cross-validation models. In/Out keep_cross_validation_fold_assignmentboolean Whether to keep the cross-validation fold assignment. In/Out parallelize_cross_validationboolean Allow parallel training of cross-validation models In/Out response_columnVecSpecifier Response variable column. In/Out weights_columnVecSpecifier Column with observation weights. Giving some observation a weight of zero is equivalent to excluding it from the dataset; giving an observation a relative weight of 2 is equivalent to repeating that row twice. Negative weights are not allowed. In/Out offset_columnVecSpecifier Offset column. This will be added to the combination of columns before applying the link function. In/Out fold_columnVecSpecifier Column with cross-validation fold index assignment per observation. In/Out fold_assignmentenum Cross-validation fold assignment scheme, if fold_column is not specified. The ‘Stratified’ option will stratify the folds based on the response variable, for classification problems. In/Out categorical_encodingenum Encoding scheme for categorical features In/Out ignored_columnsstring[] Names of columns to ignore for training. In/Out ignore_const_colsboolean Ignore constant columns. In/Out score_each_iterationboolean Whether to score during each iteration of model training. In/Out checkpointKey Model checkpoint to resume training with. In/Out stopping_roundsint Early stopping based on convergence of stopping_metric. Stop if simple moving average of length k of the stopping_metric does not improve for k:=stopping_rounds scoring events (0 to disable) In/Out max_runtime_secsdouble Maximum allowed runtime in seconds for model training. Use 0 to disable. In/Out stopping_metricenum Metric to use for early stopping (AUTO: logloss for classification, deviance for regression) In/Out stopping_tolerancedouble Relative tolerance for metric-based stopping criterion (stop if relative improvement is not at least this much) In/Out

## CoxPHV3

 parametersCoxPHParameters Model builder parameters. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In algostring The algo name for this ModelBuilder. Out algo_full_namestring The pretty algo name for this ModelBuilder (e.g., Generalized Linear Model, rather than GLM). Out can_buildenum[] Model categories this ModelBuilder can build. Out supervisedboolean Indicator whether the model is supervised or not. Out visibilityenum Should the builder always be visible, be marked as beta, or only visible if the user starts up with the experimental flag? Out jobJob Job Key Out messagesValidationMessage[] Parameter validation messages Out error_countint Count of parameter validation errors Out __http_statusint HTTP status to return for this build. Out

## CreateFrameOriginalIV4

 destKey destination key In rowsint Number of rows In colsint Number of data columns (in addition to the first response column) In seedlong Random number seed that determines the random values In randomizeboolean Whether frame should be randomized In valuelong Constant value (for randomize=false) In real_rangedouble Range for real variables (-range … range) In categorical_fractiondouble Fraction of categorical columns (for randomize=true) In factorsint Factor levels for categorical variables In integer_fractiondouble Fraction of integer columns (for randomize=true) In integer_rangeint Range for integer variables (-range … range) In binary_fractiondouble Fraction of binary columns (for randomize=true) In binary_ones_fractiondouble Fraction of 1’s in binary columns In time_fractiondouble Fraction of date/time columns (for randomize=true) In string_fractiondouble Fraction of string columns (for randomize=true) In missing_fractiondouble Fraction of missing values In has_responseboolean Whether an additional response column should be generated In response_factorsint Number of factor levels of the first column (1=real, 2=binomial, N=multinomial) In positive_responseboolean For real-valued response variable: Whether the response should be positive only. In _fieldsstring Filter on the set of output fields: if you set _fields=”foo,bar,baz”, then only those fields will be included in the output; or you can specify _fields=”-goo,gee” to include all fields except goo and gee. If the result contains nested data structures, then you can refer to the fields within those structures as well. For example if you specify _fields=”foo(oof),bar(-rab)”, then only fields foo and bar will be included, and within foo there will be only field oof, whereas within bar all fields except rab will be reported. In

## CreateFrameSimpleIV4

 destKey Id for the frame to be created. In seedlong Random number seed that determines the random values. In nrowsint Number of rows. In ncols_realint Number of real-valued columns. Values in these columns will be uniformly distributed between real_lb and real_ub. In ncols_intint Number of integer columns. In ncols_enumint Number of enum (categorical) columns. In ncols_boolint Number of boolean (binary) columns. In ncols_strint Number of string columns. In ncols_timeint Number of time columns. In real_lbdouble Lower bound for the range of the real-valued columns. In real_ubdouble Upper bound for the range of the real-valued columns. In int_lbint Lower bound for the range of integer columns. In int_ubint Upper bound for the range of integer columns. In enum_nlevelsint Number of levels (categories) for the enum columns. In bool_pdouble Fraction of ones in each boolean (binary) column. In time_lblong Lower bound for the range of time columns (in ms since the epoch). In time_ublong Upper bound for the range of time columns (in ms since the epoch). In str_lengthint Length of generated strings in string columns. In missing_fractiondouble Fraction of missing values. In response_typeenum Type of the response column to add. In response_lbdouble Lower bound for the response variable (real/int/time types). In response_ubdouble Upper bound for the response variable (real/int/time types). In response_pdouble Frequency of 1s for the bool (binary) response column. In response_nlevelsint Number of categorical levels for the enum response column. In _fieldsstring Filter on the set of output fields: if you set _fields=”foo,bar,baz”, then only those fields will be included in the output; or you can specify _fields=”-goo,gee” to include all fields except goo and gee. If the result contains nested data structures, then you can refer to the fields within those structures as well. For example if you specify _fields=”foo(oof),bar(-rab)”, then only fields foo and bar will be included, and within foo there will be only field oof, whereas within bar all fields except rab will be reported. In

## CreateFrameV3

 _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In destKey destination key In/Out rowslong Number of rows In/Out colsint Number of data columns (in addition to the first response column) In/Out seedlong Random number seed that determines the random values In/Out seed_for_column_typeslong Random number seed for setting the column types In/Out randomizeboolean Whether frame should be randomized In/Out valuelong Constant value (for randomize=false) In/Out real_rangelong Range for real variables (-range … range) In/Out categorical_fractiondouble Fraction of categorical columns (for randomize=true) In/Out factorsint Factor levels for categorical variables In/Out integer_fractiondouble Fraction of integer columns (for randomize=true) In/Out integer_rangelong Range for integer variables (-range … range) In/Out binary_fractiondouble Fraction of binary columns (for randomize=true) In/Out binary_ones_fractiondouble Fraction of 1’s in binary columns In/Out time_fractiondouble Fraction of date/time columns (for randomize=true) In/Out string_fractiondouble Fraction of string columns (for randomize=true) In/Out missing_fractiondouble Fraction of missing values In/Out has_responseboolean Whether an additional response column should be generated In/Out response_factorsint Number of factor levels of the first column (1=real, 2=binomial, N=multinomial) In/Out positive_responseboolean For real-valued response variable: Whether the response should be positive only. In/Out keyKey Job Key Out

## DCTTransformerV3

 datasetKey Dataset In destination_frameKey Destination Frame ID In dimensionsint[] Dimensions of the input array: Height, Width, Depth (Nx1x1 for 1D, NxMx1 for 2D) In inverseboolean Whether to do the inverse transform In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In

## DRFModelOutputV3

 variable_importancesTwoDimTable Variable Importances Out init_fdouble The Intercept term, the initial model function value to which trees make adjustments Out namesstring[] Column names Out domainsstring[][] Domains for categorical columns Out cross_validation_modelsKey[] Cross-validation models (model ids) Out cross_validation_predictionsKey[] Cross-validation predictions, one per cv model (deprecated, use cross_validation_holdout_predictions_frame_id instead) Out cross_validation_holdout_predictions_frame_idKey Cross-validation holdout predictions (full out-of-sample predictions on training data) Out cross_validation_fold_assignment_frame_idKey Cross-validation fold assignment (each row is assigned to one holdout fold) Out model_categoryenum Category of the model (e.g., Binomial) Out model_summaryTwoDimTable Model summary Out scoring_historyTwoDimTable Scoring history Out training_metricsModelMetrics Training data model metrics Out validation_metricsModelMetrics Validation data model metrics Out cross_validation_metricsModelMetrics Cross-validation model metrics Out cross_validation_metrics_summaryTwoDimTable Cross-validation model metrics summary Out statusstring Job status Out start_timelong Start time in milliseconds Out end_timelong End time in milliseconds Out run_timelong Runtime in milliseconds Out helpMap Help information for output fields Out

## DRFModelV3

 model_idKey Model key In/Out parametersDRFParameters The build parameters for the model (e.g. K for KMeans). Out outputDRFOutput The build output for the model (e.g. the cluster centers for KMeans). Out compatible_framesstring[] Compatible frames, if requested Out checksumlong Checksum for all the things that go into building the Model. Out algostring The algo name for this Model. Out algo_full_namestring The pretty algo name for this Model (e.g., Generalized Linear Model, rather than GLM). Out response_column_namestring The response column name for this Model (if applicable). Is null otherwise. Out data_frameKey The Model’s training frame key Out timestamplong Timestamp for when this model was completed Out

## DRFParametersV3

 mtriesint Number of variables randomly sampled as candidates at each split. If set to -1, defaults to sqrt{p} for classification and p/3 for regression (where p is the # of predictors In binomial_double_treesboolean For binary classification: Build 2x as many trees (one per class) - can lead to higher accuracy. In ntreesint Number of trees. In max_depthint Maximum tree depth. In min_rowsdouble Fewest allowed (weighted) observations in a leaf. In nbinsint For numerical columns (real/int), build a histogram of (at least) this many bins, then split at the best point In nbins_top_levelint For numerical columns (real/int), build a histogram of (at most) this many bins at the root level, then decrease by factor of two per level In nbins_catsint For categorical columns (factors), build a histogram of this many bins, then split at the best point. Higher values can lead to more overfitting. In r2_stoppingdouble r2_stopping is no longer supported and will be ignored if set - please use stopping_rounds, stopping_metric and stopping_tolerance instead. Previous version of H2O would stop making trees when the R^2 metric equals or exceeds this In seedlong Seed for pseudo random number generator (if applicable) In build_tree_one_nodeboolean Run on one node only; no network overhead but fewer cpus used. Suitable for small datasets. In sample_ratedouble Row sample rate per tree (from 0.0 to 1.0) In sample_rate_per_classdouble[] Row sample rate per tree per class (from 0.0 to 1.0) In col_sample_rate_per_treedouble Column sample rate per tree (from 0.0 to 1.0) In col_sample_rate_change_per_leveldouble Relative change of the column sampling rate for every level (from 0.0 to 2.0) In score_tree_intervalint Score the model after every so many trees. Disabled if set to 0. In min_split_improvementdouble Minimum relative improvement in squared error reduction for a split to happen In histogram_typeenum What type of histogram to use for finding optimal split points In distributionenum Distribution function In tweedie_powerdouble Tweedie power for Tweedie regression, must be between 1 and 2. In quantile_alphadouble Desired quantile for Quantile regression, must be between 0 and 1. In huber_alphadouble Desired quantile for Huber/M-regression (threshold between quadratic and linear loss, must be between 0 and 1). In balance_classesboolean Balance training data class counts via over/under-sampling (for imbalanced data). In/Out class_sampling_factorsfloat[] Desired over/under-sampling ratios per class (in lexicographic order). If not specified, sampling factors will be automatically computed to obtain class balance during training. Requires balance_classes. In/Out max_after_balance_sizefloat Maximum relative size of the training data after balancing class counts (can be less than 1.0). Requires balance_classes. In/Out max_confusion_matrix_sizeint [Deprecated] Maximum size (# classes) for confusion matrices to be printed in the Logs In/Out max_hit_ratio_kint Max. number (top K) of predictions to use for hit ratio computation (for multi-class only, 0 to disable) In/Out model_idKey Destination id for this model; auto-generated if not specified. In/Out training_frameKey Id of the training data frame (Not required, to allow initial validation of model parameters). In/Out validation_frameKey Id of the validation data frame. In/Out nfoldsint Number of folds for N-fold cross-validation (0 to disable or >= 2). In/Out keep_cross_validation_predictionsboolean Whether to keep the predictions of the cross-validation models. In/Out keep_cross_validation_fold_assignmentboolean Whether to keep the cross-validation fold assignment. In/Out parallelize_cross_validationboolean Allow parallel training of cross-validation models In/Out response_columnVecSpecifier Response variable column. In/Out weights_columnVecSpecifier Column with observation weights. Giving some observation a weight of zero is equivalent to excluding it from the dataset; giving an observation a relative weight of 2 is equivalent to repeating that row twice. Negative weights are not allowed. In/Out offset_columnVecSpecifier Offset column. This will be added to the combination of columns before applying the link function. In/Out fold_columnVecSpecifier Column with cross-validation fold index assignment per observation. In/Out fold_assignmentenum Cross-validation fold assignment scheme, if fold_column is not specified. The ‘Stratified’ option will stratify the folds based on the response variable, for classification problems. In/Out categorical_encodingenum Encoding scheme for categorical features In/Out ignored_columnsstring[] Names of columns to ignore for training. In/Out ignore_const_colsboolean Ignore constant columns. In/Out score_each_iterationboolean Whether to score during each iteration of model training. In/Out checkpointKey Model checkpoint to resume training with. In/Out stopping_roundsint Early stopping based on convergence of stopping_metric. Stop if simple moving average of length k of the stopping_metric does not improve for k:=stopping_rounds scoring events (0 to disable) In/Out max_runtime_secsdouble Maximum allowed runtime in seconds for model training. Use 0 to disable. In/Out stopping_metricenum Metric to use for early stopping (AUTO: logloss for classification, deviance for regression) In/Out stopping_tolerancedouble Relative tolerance for metric-based stopping criterion (stop if relative improvement is not at least this much) In/Out

## DRFV3

 parametersDRFParameters Model builder parameters. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In algostring The algo name for this ModelBuilder. Out algo_full_namestring The pretty algo name for this ModelBuilder (e.g., Generalized Linear Model, rather than GLM). Out can_buildenum[] Model categories this ModelBuilder can build. Out supervisedboolean Indicator whether the model is supervised or not. Out visibilityenum Should the builder always be visible, be marked as beta, or only visible if the user starts up with the experimental flag? Out jobJob Job Key Out messagesValidationMessage[] Parameter validation messages Out error_countint Count of parameter validation errors Out __http_statusint HTTP status to return for this build. Out

## DStackTraceV3

 nodestring Node name Out timelong Unix epoch time Out thread_tracesstring[] One trace per thread Out

## DataInfoFrameV3

 frameKey input frame In interactionsstring[] interactions In use_allboolean use all factor levels In standardizeboolean standardize In interactions_onlyboolean interactions only returned In resultKey output frame Out

## DeepLearningModelOutputV3

 weightsKey[] Frame keys for weight matrices In biasesKey[] Frame keys for bias vectors In normmuldouble[] Normalization/Standardization multipliers for numeric predictors Out normsubdouble[] Normalization/Standardization offsets for numeric predictors Out normrespmuldouble[] Normalization/Standardization multipliers for numeric response Out normrespsubdouble[] Normalization/Standardization offsets for numeric response Out catoffsetsint[] Categorical offsets for one-hot encoding Out variable_importancesTwoDimTable Variable Importances Out namesstring[] Column names Out domainsstring[][] Domains for categorical columns Out cross_validation_modelsKey[] Cross-validation models (model ids) Out cross_validation_predictionsKey[] Cross-validation predictions, one per cv model (deprecated, use cross_validation_holdout_predictions_frame_id instead) Out cross_validation_holdout_predictions_frame_idKey Cross-validation holdout predictions (full out-of-sample predictions on training data) Out cross_validation_fold_assignment_frame_idKey Cross-validation fold assignment (each row is assigned to one holdout fold) Out model_categoryenum Category of the model (e.g., Binomial) Out model_summaryTwoDimTable Model summary Out scoring_historyTwoDimTable Scoring history Out training_metricsModelMetrics Training data model metrics Out validation_metricsModelMetrics Validation data model metrics Out cross_validation_metricsModelMetrics Cross-validation model metrics Out cross_validation_metrics_summaryTwoDimTable Cross-validation model metrics summary Out statusstring Job status Out start_timelong Start time in milliseconds Out end_timelong End time in milliseconds Out run_timelong Runtime in milliseconds Out helpMap Help information for output fields Out

## DeepLearningModelV3

 model_idKey Model key In/Out parametersDeepLearningParameters The build parameters for the model (e.g. K for KMeans). Out outputDeepLearningModelOutput The build output for the model (e.g. the cluster centers for KMeans). Out compatible_framesstring[] Compatible frames, if requested Out checksumlong Checksum for all the things that go into building the Model. Out algostring The algo name for this Model. Out algo_full_namestring The pretty algo name for this Model (e.g., Generalized Linear Model, rather than GLM). Out response_column_namestring The response column name for this Model (if applicable). Is null otherwise. Out data_frameKey The Model’s training frame key Out timestamplong Timestamp for when this model was completed Out

## DeepLearningParametersV3

 distributionenum Distribution function In tweedie_powerdouble Tweedie power for Tweedie regression, must be between 1 and 2. In quantile_alphadouble Desired quantile for Quantile regression, must be between 0 and 1. In huber_alphadouble Desired quantile for Huber/M-regression (threshold between quadratic and linear loss, must be between 0 and 1). In balance_classesboolean Balance training data class counts via over/under-sampling (for imbalanced data). In/Out class_sampling_factorsfloat[] Desired over/under-sampling ratios per class (in lexicographic order). If not specified, sampling factors will be automatically computed to obtain class balance during training. Requires balance_classes. In/Out max_after_balance_sizefloat Maximum relative size of the training data after balancing class counts (can be less than 1.0). Requires balance_classes. In/Out max_confusion_matrix_sizeint [Deprecated] Maximum size (# classes) for confusion matrices to be printed in the Logs. In/Out max_hit_ratio_kint Max. number (top K) of predictions to use for hit ratio computation (for multi-class only, 0 to disable). In/Out activationenum Activation function. In/Out hiddenint[] Hidden layer sizes (e.g. [100, 100]). In/Out epochsdouble How many times the dataset should be iterated (streamed), can be fractional. In/Out train_samples_per_iterationlong Number of training samples (globally) per MapReduce iteration. Special values are 0: one epoch, -1: all available data (e.g., replicated training data), -2: automatic. In/Out target_ratio_comm_to_compdouble Target ratio of communication overhead to computation. Only for multi-node operation and train_samples_per_iteration = -2 (auto-tuning). In/Out seedlong Seed for random numbers (affects sampling) - Note: only reproducible when running single threaded. In/Out adaptive_rateboolean Adaptive learning rate. In/Out rhodouble Adaptive learning rate time decay factor (similarity to prior updates). In/Out epsilondouble Adaptive learning rate smoothing factor (to avoid divisions by zero and allow progress). In/Out ratedouble Learning rate (higher => less stable, lower => slower convergence). In/Out rate_annealingdouble Learning rate annealing: rate / (1 + rate_annealing * samples). In/Out rate_decaydouble Learning rate decay factor between layers (N-th layer: rate * rate_decay ^ (n - 1). In/Out momentum_startdouble Initial momentum at the beginning of training (try 0.5). In/Out momentum_rampdouble Number of training samples for which momentum increases. In/Out momentum_stabledouble Final momentum after the ramp is over (try 0.99). In/Out nesterov_accelerated_gradientboolean Use Nesterov accelerated gradient (recommended). In/Out input_dropout_ratiodouble Input layer dropout ratio (can improve generalization, try 0.1 or 0.2). In/Out hidden_dropout_ratiosdouble[] Hidden layer dropout ratios (can improve generalization), specify one value per hidden layer, defaults to 0.5. In/Out l1double L1 regularization (can add stability and improve generalization, causes many weights to become 0). In/Out l2double L2 regularization (can add stability and improve generalization, causes many weights to be small. In/Out max_w2float Constraint for squared sum of incoming weights per unit (e.g. for Rectifier). In/Out initial_weight_distributionenum Initial weight distribution. In/Out initial_weight_scaledouble Uniform: -value…value, Normal: stddev. In/Out initial_weightsKey[] A list of H2OFrame ids to initialize the weight matrices of this model with. In/Out initial_biasesKey[] A list of H2OFrame ids to initialize the bias vectors of this model with. In/Out lossenum Loss function. In/Out score_intervaldouble Shortest time interval (in seconds) between model scoring. In/Out score_training_sampleslong Number of training set samples for scoring (0 for all). In/Out score_validation_sampleslong Number of validation set samples for scoring (0 for all). In/Out score_duty_cycledouble Maximum duty cycle fraction for scoring (lower: more training, higher: more scoring). In/Out classification_stopdouble Stopping criterion for classification error fraction on training data (-1 to disable). In/Out regression_stopdouble Stopping criterion for regression error (MSE) on training data (-1 to disable). In/Out quiet_modeboolean Enable quiet mode for less output to standard output. In/Out score_validation_samplingenum Method used to sample validation dataset for scoring. In/Out overwrite_with_best_modelboolean If enabled, override the final model with the best model found during training. In/Out autoencoderboolean Auto-Encoder. In/Out use_all_factor_levelsboolean Use all factor levels of categorical variables. Otherwise, the first factor level is omitted (without loss of accuracy). Useful for variable importances and auto-enabled for autoencoder. In/Out standardizeboolean If enabled, automatically standardize the data. If disabled, the user must provide properly scaled input data. In/Out diagnosticsboolean Enable diagnostics for hidden layers. In/Out variable_importancesboolean Compute variable importances for input features (Gedeon method) - can be slow for large networks. In/Out fast_modeboolean Enable fast mode (minor approximation in back-propagation). In/Out force_load_balanceboolean Force extra load balancing to increase training speed for small datasets (to keep all cores busy). In/Out replicate_training_databoolean Replicate the entire training dataset onto every node for faster training on small datasets. In/Out single_node_modeboolean Run on a single node for fine-tuning of model parameters. In/Out shuffle_training_databoolean Enable shuffling of training data (recommended if training data is replicated and train_samples_per_iteration is close to #nodes x #rows, of if using balance_classes). In/Out missing_values_handlingenum Handling of missing values. Either MeanImputation or Skip. In/Out sparseboolean Sparse data handling (more efficient for data with lots of 0 values). In/Out col_majorboolean id="deprecated-use-a-column-major-weight-matrix-for-input-layer-can-speed-up-forward-propagation-but-might-slow-down-backpropagation-">DEPRECATED Use a column major weight matrix for input layer. Can speed up forward propagation, but might slow down backpropagation.< In/Out average_activationdouble Average activation for sparse auto-encoder. #Experimental In/Out sparsity_betadouble Sparsity regularization. #Experimental In/Out max_categorical_featuresint Max. number of categorical features, enforced via hashing. #Experimental In/Out reproducibleboolean Force reproducibility on small data (will be slow - only uses 1 thread). In/Out export_weights_and_biasesboolean Whether to export Neural Network weights and biases to H2O Frames. In/Out mini_batch_sizeint Mini-batch size (smaller leads to better fit, larger can speed up and generalize better). In/Out elastic_averagingboolean Elastic averaging between compute nodes can improve distributed model convergence. #Experimental In/Out elastic_averaging_moving_ratedouble Elastic averaging moving rate (only if elastic averaging is enabled). In/Out elastic_averaging_regularizationdouble Elastic averaging regularization strength (only if elastic averaging is enabled). In/Out pretrained_autoencoderKey Pretrained autoencoder model to initialize this model with. In/Out model_idKey Destination id for this model; auto-generated if not specified. In/Out training_frameKey Id of the training data frame (Not required, to allow initial validation of model parameters). In/Out validation_frameKey Id of the validation data frame. In/Out nfoldsint Number of folds for N-fold cross-validation (0 to disable or >= 2). In/Out keep_cross_validation_predictionsboolean Whether to keep the predictions of the cross-validation models. In/Out keep_cross_validation_fold_assignmentboolean Whether to keep the cross-validation fold assignment. In/Out parallelize_cross_validationboolean Allow parallel training of cross-validation models In/Out response_columnVecSpecifier Response variable column. In/Out weights_columnVecSpecifier Column with observation weights. Giving some observation a weight of zero is equivalent to excluding it from the dataset; giving an observation a relative weight of 2 is equivalent to repeating that row twice. Negative weights are not allowed. In/Out offset_columnVecSpecifier Offset column. This will be added to the combination of columns before applying the link function. In/Out fold_columnVecSpecifier Column with cross-validation fold index assignment per observation. In/Out fold_assignmentenum Cross-validation fold assignment scheme, if fold_column is not specified. The ‘Stratified’ option will stratify the folds based on the response variable, for classification problems. In/Out categorical_encodingenum Encoding scheme for categorical features In/Out ignored_columnsstring[] Names of columns to ignore for training. In/Out ignore_const_colsboolean Ignore constant columns. In/Out score_each_iterationboolean Whether to score during each iteration of model training. In/Out checkpointKey Model checkpoint to resume training with. In/Out stopping_roundsint Early stopping based on convergence of stopping_metric. Stop if simple moving average of length k of the stopping_metric does not improve for k:=stopping_rounds scoring events (0 to disable) In/Out max_runtime_secsdouble Maximum allowed runtime in seconds for model training. Use 0 to disable. In/Out stopping_metricenum Metric to use for early stopping (AUTO: logloss for classification, deviance for regression) In/Out stopping_tolerancedouble Relative tolerance for metric-based stopping criterion (stop if relative improvement is not at least this much) In/Out

## DeepLearningV3

 parametersDeepLearningParameters Model builder parameters. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In algostring The algo name for this ModelBuilder. Out algo_full_namestring The pretty algo name for this ModelBuilder (e.g., Generalized Linear Model, rather than GLM). Out can_buildenum[] Model categories this ModelBuilder can build. Out supervisedboolean Indicator whether the model is supervised or not. Out visibilityenum Should the builder always be visible, be marked as beta, or only visible if the user starts up with the experimental flag? Out jobJob Job Key Out messagesValidationMessage[] Parameter validation messages Out error_countint Count of parameter validation errors Out __http_statusint HTTP status to return for this build. Out

## DeepWaterModelOutputV3

 namesstring[] Column names Out domainsstring[][] Domains for categorical columns Out cross_validation_modelsKey[] Cross-validation models (model ids) Out cross_validation_predictionsKey[] Cross-validation predictions, one per cv model (deprecated, use cross_validation_holdout_predictions_frame_id instead) Out cross_validation_holdout_predictions_frame_idKey Cross-validation holdout predictions (full out-of-sample predictions on training data) Out cross_validation_fold_assignment_frame_idKey Cross-validation fold assignment (each row is assigned to one holdout fold) Out model_categoryenum Category of the model (e.g., Binomial) Out model_summaryTwoDimTable Model summary Out scoring_historyTwoDimTable Scoring history Out training_metricsModelMetrics Training data model metrics Out validation_metricsModelMetrics Validation data model metrics Out cross_validation_metricsModelMetrics Cross-validation model metrics Out cross_validation_metrics_summaryTwoDimTable Cross-validation model metrics summary Out statusstring Job status Out start_timelong Start time in milliseconds Out end_timelong End time in milliseconds Out run_timelong Runtime in milliseconds Out helpMap Help information for output fields Out

## DeepWaterModelV3

 model_idKey Model key In/Out parametersDeepWaterParameters The build parameters for the model (e.g. K for KMeans). Out outputDeepWaterModelOutput The build output for the model (e.g. the cluster centers for KMeans). Out compatible_framesstring[] Compatible frames, if requested Out checksumlong Checksum for all the things that go into building the Model. Out algostring The algo name for this Model. Out algo_full_namestring The pretty algo name for this Model (e.g., Generalized Linear Model, rather than GLM). Out response_column_namestring The response column name for this Model (if applicable). Is null otherwise. Out data_frameKey The Model’s training frame key Out timestamplong Timestamp for when this model was completed Out

## DeepWaterParametersV3

 distributionenum Distribution function In tweedie_powerdouble Tweedie power for Tweedie regression, must be between 1 and 2. In quantile_alphadouble Desired quantile for Quantile regression, must be between 0 and 1. In huber_alphadouble Desired quantile for Huber/M-regression (threshold between quadratic and linear loss, must be between 0 and 1). In problem_typeenum Problem type, auto-detected by default. If set to image, the H2OFrame must contain a string column containing the path (URI or URL) to the images in the first column. If set to text, the H2OFrame must contain a string column containing the text in the first column. If set to dataset, Deep Water behaves just like any other H2O Model and builds a model on the provided H2OFrame (non-String columns). In/Out activationenum Activation function. Only used if no user-defined network architecture file is provided, and only for problem_type=dataset. In/Out hiddenint[] Hidden layer sizes (e.g. [200, 200]). Only used if no user-defined network architecture file is provided, and only for problem_type=dataset. In/Out input_dropout_ratiodouble Input layer dropout ratio (can improve generalization, try 0.1 or 0.2). In/Out hidden_dropout_ratiosdouble[] Hidden layer dropout ratios (can improve generalization), specify one value per hidden layer, defaults to 0.5. In/Out max_confusion_matrix_sizeint [Deprecated] Maximum size (# classes) for confusion matrices to be printed in the Logs. In/Out sparseboolean Sparse data handling (more efficient for data with lots of 0 values). In/Out max_hit_ratio_kint Max. number (top K) of predictions to use for hit ratio computation (for multi-class only, 0 to disable). In/Out epochsdouble How many times the dataset should be iterated (streamed), can be fractional. In/Out train_samples_per_iterationlong Number of training samples (globally) per MapReduce iteration. Special values are 0: one epoch, -1: all available data (e.g., replicated training data), -2: automatic. In/Out target_ratio_comm_to_compdouble Target ratio of communication overhead to computation. Only for multi-node operation and train_samples_per_iteration = -2 (auto-tuning). In/Out seedlong Seed for random numbers (affects sampling) - Note: only reproducible when running single threaded. In/Out learning_ratedouble Learning rate (higher => less stable, lower => slower convergence). In/Out learning_rate_annealingdouble Learning rate annealing: rate / (1 + rate_annealing * samples). In/Out momentum_startdouble Initial momentum at the beginning of training (try 0.5). In/Out momentum_rampdouble Number of training samples for which momentum increases. In/Out momentum_stabledouble Final momentum after the ramp is over (try 0.99). In/Out score_intervaldouble Shortest time interval (in seconds) between model scoring. In/Out score_training_sampleslong Number of training set samples for scoring (0 for all). In/Out score_validation_sampleslong Number of validation set samples for scoring (0 for all). In/Out score_duty_cycledouble Maximum duty cycle fraction for scoring (lower: more training, higher: more scoring). In/Out classification_stopdouble Stopping criterion for classification error fraction on training data (-1 to disable). In/Out regression_stopdouble Stopping criterion for regression error (MSE) on training data (-1 to disable). In/Out quiet_modeboolean Enable quiet mode for less output to standard output. In/Out overwrite_with_best_modelboolean If enabled, override the final model with the best model found during training. In/Out autoencoderboolean Auto-Encoder. In/Out diagnosticsboolean Enable diagnostics for hidden layers. In/Out variable_importancesboolean Compute variable importances for input features (Gedeon method) - can be slow for large networks. In/Out replicate_training_databoolean Replicate the entire training dataset onto every node for faster training on small datasets. In/Out single_node_modeboolean Run on a single node for fine-tuning of model parameters. In/Out shuffle_training_databoolean Enable global shuffling of training data. In/Out mini_batch_sizeint Mini-batch size (smaller leads to better fit, larger can speed up and generalize better). In/Out clip_gradientdouble Clip gradients once their absolute value is larger than this value. In/Out networkenum Network architecture. In/Out backendenum Deep Learning Backend. In/Out image_shapeint[] Width and height of image. In/Out channelsint Number of (color) channels. In/Out gpuboolean Whether to use a GPU (if available). In/Out device_idint[] Device IDs (which GPUs to use). In/Out network_definition_filestring Path of file containing network definition (graph, architecture). In/Out network_parameters_filestring Path of file containing network (initial) parameters (weights, biases). In/Out mean_image_filestring Path of file containing the mean image data for data normalization. In/Out export_native_parameters_prefixstring Path (prefix) where to export the native model parameters after every iteration. In/Out standardizeboolean If enabled, automatically standardize the data. If disabled, the user must provide properly scaled input data. In/Out balance_classesboolean Balance training data class counts via over/under-sampling (for imbalanced data). In/Out class_sampling_factorsfloat[] Desired over/under-sampling ratios per class (in lexicographic order). If not specified, sampling factors will be automatically computed to obtain class balance during training. Requires balance_classes. In/Out max_after_balance_sizefloat Maximum relative size of the training data after balancing class counts (can be less than 1.0). Requires balance_classes. In/Out model_idKey Destination id for this model; auto-generated if not specified. In/Out training_frameKey Id of the training data frame (Not required, to allow initial validation of model parameters). In/Out validation_frameKey Id of the validation data frame. In/Out nfoldsint Number of folds for N-fold cross-validation (0 to disable or >= 2). In/Out keep_cross_validation_predictionsboolean Whether to keep the predictions of the cross-validation models. In/Out keep_cross_validation_fold_assignmentboolean Whether to keep the cross-validation fold assignment. In/Out parallelize_cross_validationboolean Allow parallel training of cross-validation models In/Out response_columnVecSpecifier Response variable column. In/Out weights_columnVecSpecifier Column with observation weights. Giving some observation a weight of zero is equivalent to excluding it from the dataset; giving an observation a relative weight of 2 is equivalent to repeating that row twice. Negative weights are not allowed. In/Out offset_columnVecSpecifier Offset column. This will be added to the combination of columns before applying the link function. In/Out fold_columnVecSpecifier Column with cross-validation fold index assignment per observation. In/Out fold_assignmentenum Cross-validation fold assignment scheme, if fold_column is not specified. The ‘Stratified’ option will stratify the folds based on the response variable, for classification problems. In/Out categorical_encodingenum Encoding scheme for categorical features In/Out ignored_columnsstring[] Names of columns to ignore for training. In/Out ignore_const_colsboolean Ignore constant columns. In/Out score_each_iterationboolean Whether to score during each iteration of model training. In/Out checkpointKey Model checkpoint to resume training with. In/Out stopping_roundsint Early stopping based on convergence of stopping_metric. Stop if simple moving average of length k of the stopping_metric does not improve for k:=stopping_rounds scoring events (0 to disable) In/Out max_runtime_secsdouble Maximum allowed runtime in seconds for model training. Use 0 to disable. In/Out stopping_metricenum Metric to use for early stopping (AUTO: logloss for classification, deviance for regression) In/Out stopping_tolerancedouble Relative tolerance for metric-based stopping criterion (stop if relative improvement is not at least this much) In/Out

## DeepWaterV3

 parametersDeepWaterParameters Model builder parameters. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In algostring The algo name for this ModelBuilder. Out algo_full_namestring The pretty algo name for this ModelBuilder (e.g., Generalized Linear Model, rather than GLM). Out can_buildenum[] Model categories this ModelBuilder can build. Out supervisedboolean Indicator whether the model is supervised or not. Out visibilityenum Should the builder always be visible, be marked as beta, or only visible if the user starts up with the experimental flag? Out jobJob Job Key Out messagesValidationMessage[] Parameter validation messages Out error_countint Count of parameter validation errors Out __http_statusint HTTP status to return for this build. Out

 frame_idKey Frame to download In hex_stringboolean Emit double values in a machine readable lossless format with Double.toHexString(). In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In csvstring CSV Stream Out filenamestring Suggested Filename Out

## EndpointV4

 urlstring Method+Url of the request; variable parts are enclosed in curly braces. For example: /4/schemas/{schema_name} In descriptionstring Short description of the functionality provided by the endpoint. In namestring Unique name of the endpoint. These names can be used to look up the endpoint’s info via GET /4/endpoints/{name}. In input_schemastring Input schema. In output_schemastring Schema for the result returned by the endpoint. In __schemastring Url describing the schema of the current object. In

## EndpointsListV4

 endpointsRoute[] List of endpoints in H2O REST API (v4). In __schemastring Url describing the schema of the current object. In

## EventV3

 datestring Time when the event was recorded. Format is hh:mm:ss:ms In nanoslong Time in nanos In typeenum type of recorded event In

## ExampleModelOutputV3

 iterationsint Iterations executed In maxsdouble[] (No description available) In namesstring[] Column names Out domainsstring[][] Domains for categorical columns Out cross_validation_modelsKey[] Cross-validation models (model ids) Out cross_validation_predictionsKey[] Cross-validation predictions, one per cv model (deprecated, use cross_validation_holdout_predictions_frame_id instead) Out cross_validation_holdout_predictions_frame_idKey Cross-validation holdout predictions (full out-of-sample predictions on training data) Out cross_validation_fold_assignment_frame_idKey Cross-validation fold assignment (each row is assigned to one holdout fold) Out model_categoryenum Category of the model (e.g., Binomial) Out model_summaryTwoDimTable Model summary Out scoring_historyTwoDimTable Scoring history Out training_metricsModelMetrics Training data model metrics Out validation_metricsModelMetrics Validation data model metrics Out cross_validation_metricsModelMetrics Cross-validation model metrics Out cross_validation_metrics_summaryTwoDimTable Cross-validation model metrics summary Out statusstring Job status Out start_timelong Start time in milliseconds Out end_timelong End time in milliseconds Out run_timelong Runtime in milliseconds Out helpMap Help information for output fields Out

## ExampleModelV3

 model_idKey Model key In/Out parametersExampleParameters The build parameters for the model (e.g. K for KMeans). Out outputExampleOutput The build output for the model (e.g. the cluster centers for KMeans). Out compatible_framesstring[] Compatible frames, if requested Out checksumlong Checksum for all the things that go into building the Model. Out algostring The algo name for this Model. Out algo_full_namestring The pretty algo name for this Model (e.g., Generalized Linear Model, rather than GLM). Out response_column_namestring The response column name for this Model (if applicable). Is null otherwise. Out data_frameKey The Model’s training frame key Out timestamplong Timestamp for when this model was completed Out

## ExampleParametersV3

 max_iterationsint Maximum training iterations. In distributionenum Distribution function In tweedie_powerdouble Tweedie power for Tweedie regression, must be between 1 and 2. In quantile_alphadouble Desired quantile for Quantile regression, must be between 0 and 1. In huber_alphadouble Desired quantile for Huber/M-regression (threshold between quadratic and linear loss, must be between 0 and 1). In model_idKey Destination id for this model; auto-generated if not specified. In/Out training_frameKey Id of the training data frame (Not required, to allow initial validation of model parameters). In/Out validation_frameKey Id of the validation data frame. In/Out nfoldsint Number of folds for N-fold cross-validation (0 to disable or >= 2). In/Out keep_cross_validation_predictionsboolean Whether to keep the predictions of the cross-validation models. In/Out keep_cross_validation_fold_assignmentboolean Whether to keep the cross-validation fold assignment. In/Out parallelize_cross_validationboolean Allow parallel training of cross-validation models In/Out response_columnVecSpecifier Response variable column. In/Out weights_columnVecSpecifier Column with observation weights. Giving some observation a weight of zero is equivalent to excluding it from the dataset; giving an observation a relative weight of 2 is equivalent to repeating that row twice. Negative weights are not allowed. In/Out offset_columnVecSpecifier Offset column. This will be added to the combination of columns before applying the link function. In/Out fold_columnVecSpecifier Column with cross-validation fold index assignment per observation. In/Out fold_assignmentenum Cross-validation fold assignment scheme, if fold_column is not specified. The ‘Stratified’ option will stratify the folds based on the response variable, for classification problems. In/Out categorical_encodingenum Encoding scheme for categorical features In/Out ignored_columnsstring[] Names of columns to ignore for training. In/Out ignore_const_colsboolean Ignore constant columns. In/Out score_each_iterationboolean Whether to score during each iteration of model training. In/Out checkpointKey Model checkpoint to resume training with. In/Out stopping_roundsint Early stopping based on convergence of stopping_metric. Stop if simple moving average of length k of the stopping_metric does not improve for k:=stopping_rounds scoring events (0 to disable) In/Out max_runtime_secsdouble Maximum allowed runtime in seconds for model training. Use 0 to disable. In/Out stopping_metricenum Metric to use for early stopping (AUTO: logloss for classification, deviance for regression) In/Out stopping_tolerancedouble Relative tolerance for metric-based stopping criterion (stop if relative improvement is not at least this much) In/Out

## ExampleV3

 parametersExampleParameters Model builder parameters. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In algostring The algo name for this ModelBuilder. Out algo_full_namestring The pretty algo name for this ModelBuilder (e.g., Generalized Linear Model, rather than GLM). Out can_buildenum[] Model categories this ModelBuilder can build. Out supervisedboolean Indicator whether the model is supervised or not. Out visibilityenum Should the builder always be visible, be marked as beta, or only visible if the user starts up with the experimental flag? Out jobJob Job Key Out messagesValidationMessage[] Parameter validation messages Out error_countint Count of parameter validation errors Out __http_statusint HTTP status to return for this build. Out

 schema_namestring Schema name for this field, if it is_schema, or the name of the enum, if it’s an enum. In namestring Field name in the Schema Out typestring Type for this field Out is_schemaboolean Type for this field is itself a Schema. Out valuePolymorphic Value for this field Out helpstring A short help description to appear alongside the field in a UI Out labelstring The label that should be displayed for the field if the name is insufficient Out requiredboolean Is this field required, or is the default value generally sufficient? Out levelenum How important is this field? The web UI uses the level to do a slow reveal of the parameters Out directionenum Is this field an input, output or inout? Out is_inheritedboolean Is the field inherited from the parent schema? Out inherited_fromstring If this field is inherited from a class higher in the hierarchy which one? Out is_gridableboolean Is the field gridable (i.e., it can be used in grid call) Out valuesstring[] For enum-type fields the allowed values are specified using the values annotation; this is used in UIs to tell the user the allowed values, and for validation Out jsonboolean Should this field be rendered in the JSON representation? Out is_member_of_framesstring[] For Vec-type fields this is the set of other Vec-type fields which must contain mutually exclusive values; for example, for a SupervisedModel the response_column must be mutually exclusive with the weights_column Out is_mutually_exclusive_withstring[] For Vec-type fields this is the set of Frame-type fields which must contain the named column; for example, for a SupervisedModel the response_column must be in both the training_frame and (if it’s set) the validation_frame Out

## FindV3

 keyFrame Frame to search In columnstring Column, or null for all In rowlong Starting row for search In matchstring Value to search for; leave blank for a search for missing values In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In prevlong previous row with matching value, or -1 Out nextlong next row with matching value, or -1 Out

## FrameBaseV3

 _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In frame_idKey Frame ID In/Out byte_sizelong Total data size in bytes Out is_textboolean Is this Frame raw unparsed data? Out

## FrameKeyV3

 namestring Name (string representation) for this Key. In/Out typestring Name (string representation) for the type of Keyed this Key points to. In/Out URLstring URL for the resource that this Key points to, if one exists. In/Out

## FrameSynopsisV3

 _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In frame_idKey Frame ID In/Out rowslong Number of rows in the Frame Out columnslong Number of columns in the Frame Out byte_sizelong Total data size in bytes Out is_textboolean Is this Frame raw unparsed data? Out

## FrameV3

 row_offsetlong Row offset to display In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In row_countint Number of rows to display In/Out column_offsetint Column offset to return In/Out column_countint Number of columns to return In/Out total_column_countint Total number of columns in the Frame In/Out frame_idKey Frame ID In/Out checksumlong checksum Out rowslong Number of rows in the Frame Out num_columnslong Number of columns in the Frame Out default_percentilesdouble[] Default percentiles, from 0 to 1 Out columnsVec[] Columns in the Frame Out compatible_modelsstring[] Compatible models, if requested Out chunk_summaryTwoDimTable Chunk summary Out distribution_summaryTwoDimTable Distribution summary Out byte_sizelong Total data size in bytes Out is_textboolean Is this Frame raw unparsed data? Out

## FramesV3

 frame_idKey Name of Frame of interest In columnstring Name of column of interest In find_compatible_modelsboolean Find and return compatible models? In pathstring File output path In forceboolean Overwrite existing file In num_partsint Number of part files to use (1=single file,-1=automatic) In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In row_offsetlong Row offset to return In/Out row_countint Number of rows to return In/Out column_offsetint Column offset to return In/Out column_countint Number of columns to return In/Out jobJob Job for export file Out framesIced[] Frames Out compatible_modelsModel[] Compatible models Out domainstring[][] Domains Out

## GBMModelOutputV3

 variable_importancesTwoDimTable Variable Importances Out init_fdouble The Intercept term, the initial model function value to which trees make adjustments Out namesstring[] Column names Out domainsstring[][] Domains for categorical columns Out cross_validation_modelsKey[] Cross-validation models (model ids) Out cross_validation_predictionsKey[] Cross-validation predictions, one per cv model (deprecated, use cross_validation_holdout_predictions_frame_id instead) Out cross_validation_holdout_predictions_frame_idKey Cross-validation holdout predictions (full out-of-sample predictions on training data) Out cross_validation_fold_assignment_frame_idKey Cross-validation fold assignment (each row is assigned to one holdout fold) Out model_categoryenum Category of the model (e.g., Binomial) Out model_summaryTwoDimTable Model summary Out scoring_historyTwoDimTable Scoring history Out training_metricsModelMetrics Training data model metrics Out validation_metricsModelMetrics Validation data model metrics Out cross_validation_metricsModelMetrics Cross-validation model metrics Out cross_validation_metrics_summaryTwoDimTable Cross-validation model metrics summary Out statusstring Job status Out start_timelong Start time in milliseconds Out end_timelong End time in milliseconds Out run_timelong Runtime in milliseconds Out helpMap Help information for output fields Out

## GBMModelV3

 model_idKey Model key In/Out parametersGBMParameters The build parameters for the model (e.g. K for KMeans). Out outputGBMOutput The build output for the model (e.g. the cluster centers for KMeans). Out compatible_framesstring[] Compatible frames, if requested Out checksumlong Checksum for all the things that go into building the Model. Out algostring The algo name for this Model. Out algo_full_namestring The pretty algo name for this Model (e.g., Generalized Linear Model, rather than GLM). Out response_column_namestring The response column name for this Model (if applicable). Is null otherwise. Out data_frameKey The Model’s training frame key Out timestamplong Timestamp for when this model was completed Out

## GBMParametersV3

 learn_ratedouble Learning rate (from 0.0 to 1.0) In learn_rate_annealingdouble Scale the learning rate by this factor after each tree (e.g., 0.99 or 0.999) In col_sample_ratedouble Column sample rate (from 0.0 to 1.0) In max_abs_leafnode_preddouble Maximum absolute value of a leaf node prediction In pred_noise_bandwidthdouble Bandwidth (sigma) of Gaussian multiplicative noise ~N(1,sigma) for tree node predictions In ntreesint Number of trees. In max_depthint Maximum tree depth. In min_rowsdouble Fewest allowed (weighted) observations in a leaf. In nbinsint For numerical columns (real/int), build a histogram of (at least) this many bins, then split at the best point In nbins_top_levelint For numerical columns (real/int), build a histogram of (at most) this many bins at the root level, then decrease by factor of two per level In nbins_catsint For categorical columns (factors), build a histogram of this many bins, then split at the best point. Higher values can lead to more overfitting. In r2_stoppingdouble r2_stopping is no longer supported and will be ignored if set - please use stopping_rounds, stopping_metric and stopping_tolerance instead. Previous version of H2O would stop making trees when the R^2 metric equals or exceeds this In seedlong Seed for pseudo random number generator (if applicable) In build_tree_one_nodeboolean Run on one node only; no network overhead but fewer cpus used. Suitable for small datasets. In sample_ratedouble Row sample rate per tree (from 0.0 to 1.0) In sample_rate_per_classdouble[] Row sample rate per tree per class (from 0.0 to 1.0) In col_sample_rate_per_treedouble Column sample rate per tree (from 0.0 to 1.0) In col_sample_rate_change_per_leveldouble Relative change of the column sampling rate for every level (from 0.0 to 2.0) In score_tree_intervalint Score the model after every so many trees. Disabled if set to 0. In min_split_improvementdouble Minimum relative improvement in squared error reduction for a split to happen In histogram_typeenum What type of histogram to use for finding optimal split points In distributionenum Distribution function In tweedie_powerdouble Tweedie power for Tweedie regression, must be between 1 and 2. In quantile_alphadouble Desired quantile for Quantile regression, must be between 0 and 1. In huber_alphadouble Desired quantile for Huber/M-regression (threshold between quadratic and linear loss, must be between 0 and 1). In balance_classesboolean Balance training data class counts via over/under-sampling (for imbalanced data). In/Out class_sampling_factorsfloat[] Desired over/under-sampling ratios per class (in lexicographic order). If not specified, sampling factors will be automatically computed to obtain class balance during training. Requires balance_classes. In/Out max_after_balance_sizefloat Maximum relative size of the training data after balancing class counts (can be less than 1.0). Requires balance_classes. In/Out max_confusion_matrix_sizeint [Deprecated] Maximum size (# classes) for confusion matrices to be printed in the Logs In/Out max_hit_ratio_kint Max. number (top K) of predictions to use for hit ratio computation (for multi-class only, 0 to disable) In/Out model_idKey Destination id for this model; auto-generated if not specified. In/Out training_frameKey Id of the training data frame (Not required, to allow initial validation of model parameters). In/Out validation_frameKey Id of the validation data frame. In/Out nfoldsint Number of folds for N-fold cross-validation (0 to disable or >= 2). In/Out keep_cross_validation_predictionsboolean Whether to keep the predictions of the cross-validation models. In/Out keep_cross_validation_fold_assignmentboolean Whether to keep the cross-validation fold assignment. In/Out parallelize_cross_validationboolean Allow parallel training of cross-validation models In/Out response_columnVecSpecifier Response variable column. In/Out weights_columnVecSpecifier Column with observation weights. Giving some observation a weight of zero is equivalent to excluding it from the dataset; giving an observation a relative weight of 2 is equivalent to repeating that row twice. Negative weights are not allowed. In/Out offset_columnVecSpecifier Offset column. This will be added to the combination of columns before applying the link function. In/Out fold_columnVecSpecifier Column with cross-validation fold index assignment per observation. In/Out fold_assignmentenum Cross-validation fold assignment scheme, if fold_column is not specified. The ‘Stratified’ option will stratify the folds based on the response variable, for classification problems. In/Out categorical_encodingenum Encoding scheme for categorical features In/Out ignored_columnsstring[] Names of columns to ignore for training. In/Out ignore_const_colsboolean Ignore constant columns. In/Out score_each_iterationboolean Whether to score during each iteration of model training. In/Out checkpointKey Model checkpoint to resume training with. In/Out stopping_roundsint Early stopping based on convergence of stopping_metric. Stop if simple moving average of length k of the stopping_metric does not improve for k:=stopping_rounds scoring events (0 to disable) In/Out max_runtime_secsdouble Maximum allowed runtime in seconds for model training. Use 0 to disable. In/Out stopping_metricenum Metric to use for early stopping (AUTO: logloss for classification, deviance for regression) In/Out stopping_tolerancedouble Relative tolerance for metric-based stopping criterion (stop if relative improvement is not at least this much) In/Out

## GBMV3

 parametersGBMParameters Model builder parameters. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In algostring The algo name for this ModelBuilder. Out algo_full_namestring The pretty algo name for this ModelBuilder (e.g., Generalized Linear Model, rather than GLM). Out can_buildenum[] Model categories this ModelBuilder can build. Out supervisedboolean Indicator whether the model is supervised or not. Out visibilityenum Should the builder always be visible, be marked as beta, or only visible if the user starts up with the experimental flag? Out jobJob Job Key Out messagesValidationMessage[] Parameter validation messages Out error_countint Count of parameter validation errors Out __http_statusint HTTP status to return for this build. Out

## GLMModelOutputV3

 coefficients_tableTwoDimTable Table of Coefficients In standardized_coefficient_magnitudesTwoDimTable Standardized Coefficient Magnitudes In lambda_bestdouble Lambda minimizing the objective value, only applicable with lambd search In lambda_1sedouble Lambda best + 1 standard error. Only applicable with lambda search and cross-validation In namesstring[] Column names Out domainsstring[][] Domains for categorical columns Out cross_validation_modelsKey[] Cross-validation models (model ids) Out cross_validation_predictionsKey[] Cross-validation predictions, one per cv model (deprecated, use cross_validation_holdout_predictions_frame_id instead) Out cross_validation_holdout_predictions_frame_idKey Cross-validation holdout predictions (full out-of-sample predictions on training data) Out cross_validation_fold_assignment_frame_idKey Cross-validation fold assignment (each row is assigned to one holdout fold) Out model_categoryenum Category of the model (e.g., Binomial) Out model_summaryTwoDimTable Model summary Out scoring_historyTwoDimTable Scoring history Out training_metricsModelMetrics Training data model metrics Out validation_metricsModelMetrics Validation data model metrics Out cross_validation_metricsModelMetrics Cross-validation model metrics Out cross_validation_metrics_summaryTwoDimTable Cross-validation model metrics summary Out statusstring Job status Out start_timelong Start time in milliseconds Out end_timelong End time in milliseconds Out run_timelong Runtime in milliseconds Out helpMap Help information for output fields Out

## GLMModelV3

 model_idKey Model key In/Out parametersGLMParameters The build parameters for the model (e.g. K for KMeans). Out outputGLMOutput The build output for the model (e.g. the cluster centers for KMeans). Out compatible_framesstring[] Compatible frames, if requested Out checksumlong Checksum for all the things that go into building the Model. Out algostring The algo name for this Model. Out algo_full_namestring The pretty algo name for this Model (e.g., Generalized Linear Model, rather than GLM). Out response_column_namestring The response column name for this Model (if applicable). Is null otherwise. Out data_frameKey The Model’s training frame key Out timestamplong Timestamp for when this model was completed Out

## GLMParametersV3

 seedlong Seed for pseudo random number generator (if applicable) In familyenum Family. Use binomial for classification with logistic regression, others are for regression problems. In tweedie_variance_powerdouble Tweedie variance power In tweedie_link_powerdouble Tweedie link power In solverenum AUTO will set the solver based on given data and the other parameters. IRLSM is fast on on problems with small number of predictors and for lambda-search with L1 penalty, L_BFGS scales better for datasets with many columns. Coordinate descent is experimental (beta). In alphadouble[] distribution of regularization between L1 and L2. Default value of alpha is 0 when SOLVER = ‘L-BFGS’, 0.5 otherwise In lambdadouble[] regularization strength In lambda_searchboolean use lambda search starting at lambda max, given lambda is then interpreted as lambda min In early_stoppingboolean stop early when there is no more relative improvement on train or validation (if provided) In nlambdasint Number of lambdas to be used in a search. Default indicates: If alpha is zero, with lambda search set to True, the value of nlamdas is set to 30 (fewer lambdas are needed for ridge regression) otherwise it is set to 100. In standardizeboolean Standardize numeric columns to have zero mean and unit variance In non_negativeboolean Restrict coefficients (not intercept) to be non-negative In max_iterationsint Maximum number of iterations In beta_epsilondouble converge if beta changes less (using L-infinity norm) than beta esilon, ONLY applies to IRLSM solver In objective_epsilondouble Converge if objective value changes less than this. Default indicates: If lambda_search is set to True the value of objective_epsilon is set to .0001. If the lambda_search is set to False and lambda is equal to zero, the value of objective_epsilon is set to .000001, for any other value of lambda the default value of objective_epsilon is set to .0001. In gradient_epsilondouble Converge if objective changes less (using L-infinity norm) than this, ONLY applies to L-BFGS solver. Default indicates: If lambda_search is set to False and lambda is equal to zero, the default value of gradient_epsilon is equal to .000001, otherwise the default value is .0001. If lambda_search is set to True, the conditional values above are 1E-8 and 1E-6 respectively. In obj_regdouble likelihood divider in objective value computation, default is 1/nobs In linkenum (No description available) In interceptboolean include constant term in the model In priordouble prior probability for y==1. To be used only for logistic regression iff the data has been sampled and the mean of response does not reflect reality. In lambda_min_ratiodouble Min lambda used in lambda search, specified as a ratio of lambda_max. Default indicates: if the number of observations is greater than the number of variables then lambda_min_ratio is set to 0.0001; if the number of observations is less than the number of variables then lambda_min_ratio is set to 0.01. In beta_constraintsKey beta constraints In max_active_predictorsint Maximum number of active predictors during computation. Use as a stopping criterion to prevent expensive model building with many predictors. Default indicates: If the IRLSM solver is used, the value of max_active_predictors is set to 7000 otherwise it is set to 100000000. In interactionsstring[] A list of predictor column indices to interact. All pairwise combinations will be computed for the list. In compute_p_valuesboolean request p-values computation, p-values work only with IRLSM solver and no regularization In remove_collinear_columnsboolean in case of linearly dependent columns remove some of the dependent columns In distributionenum Distribution function In tweedie_powerdouble Tweedie power for Tweedie regression, must be between 1 and 2. In quantile_alphadouble Desired quantile for Quantile regression, must be between 0 and 1. In huber_alphadouble Desired quantile for Huber/M-regression (threshold between quadratic and linear loss, must be between 0 and 1). In missing_values_handlingenum Handling of missing values. Either MeanImputation or Skip. In/Out balance_classesboolean Balance training data class counts via over/under-sampling (for imbalanced data). In/Out class_sampling_factorsfloat[] Desired over/under-sampling ratios per class (in lexicographic order). If not specified, sampling factors will be automatically computed to obtain class balance during training. Requires balance_classes. In/Out max_after_balance_sizefloat Maximum relative size of the training data after balancing class counts (can be less than 1.0). Requires balance_classes. In/Out max_confusion_matrix_sizeint [Deprecated] Maximum size (# classes) for confusion matrices to be printed in the Logs In/Out max_hit_ratio_kint Max. number (top K) of predictions to use for hit ratio computation (for multi-class only, 0 to disable) In/Out model_idKey Destination id for this model; auto-generated if not specified. In/Out training_frameKey Id of the training data frame (Not required, to allow initial validation of model parameters). In/Out validation_frameKey Id of the validation data frame. In/Out nfoldsint Number of folds for N-fold cross-validation (0 to disable or >= 2). In/Out keep_cross_validation_predictionsboolean Whether to keep the predictions of the cross-validation models. In/Out keep_cross_validation_fold_assignmentboolean Whether to keep the cross-validation fold assignment. In/Out parallelize_cross_validationboolean Allow parallel training of cross-validation models In/Out response_columnVecSpecifier Response variable column. In/Out weights_columnVecSpecifier Column with observation weights. Giving some observation a weight of zero is equivalent to excluding it from the dataset; giving an observation a relative weight of 2 is equivalent to repeating that row twice. Negative weights are not allowed. In/Out offset_columnVecSpecifier Offset column. This will be added to the combination of columns before applying the link function. In/Out fold_columnVecSpecifier Column with cross-validation fold index assignment per observation. In/Out fold_assignmentenum Cross-validation fold assignment scheme, if fold_column is not specified. The ‘Stratified’ option will stratify the folds based on the response variable, for classification problems. In/Out categorical_encodingenum Encoding scheme for categorical features In/Out ignored_columnsstring[] Names of columns to ignore for training. In/Out ignore_const_colsboolean Ignore constant columns. In/Out score_each_iterationboolean Whether to score during each iteration of model training. In/Out checkpointKey Model checkpoint to resume training with. In/Out stopping_roundsint Early stopping based on convergence of stopping_metric. Stop if simple moving average of length k of the stopping_metric does not improve for k:=stopping_rounds scoring events (0 to disable) In/Out max_runtime_secsdouble Maximum allowed runtime in seconds for model training. Use 0 to disable. In/Out stopping_metricenum Metric to use for early stopping (AUTO: logloss for classification, deviance for regression) In/Out stopping_tolerancedouble Relative tolerance for metric-based stopping criterion (stop if relative improvement is not at least this much) In/Out

## GLMRegularizationPathV3

 modelKey source model In lambdasdouble[] Computed lambda values In explained_deviance_traindouble[] explained deviance on the training set In explained_deviance_validdouble[] explained deviance on the validation set In coefficientsdouble[][] coefficients for all lambdas In coefficients_stddouble[][] standardized coefficients for all lambdas In coefficient_namesstring[] coefficient names In

## GLMV3

 parametersGLMParameters Model builder parameters. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In algostring The algo name for this ModelBuilder. Out algo_full_namestring The pretty algo name for this ModelBuilder (e.g., Generalized Linear Model, rather than GLM). Out can_buildenum[] Model categories this ModelBuilder can build. Out supervisedboolean Indicator whether the model is supervised or not. Out visibilityenum Should the builder always be visible, be marked as beta, or only visible if the user starts up with the experimental flag? Out jobJob Job Key Out messagesValidationMessage[] Parameter validation messages Out error_countint Count of parameter validation errors Out __http_statusint HTTP status to return for this build. Out

## GLRMModelOutputV3

 iterationsint Number of iterations executed In updatesint Number of updates executed In objectivedouble Current value of the objective function In avg_change_objdouble Average change in objective value on final iteration In step_sizedouble Final step size In archetypesTwoDimTable Mapping from lower dimensional k-space to training features (Y) In singular_valsdouble[] Singular values of XY matrix In eigenvectorsTwoDimTable Eigenvectors of XY matrix In representation_namestring Frame key name for X matrix In importanceTwoDimTable Standard deviation and importance of each principal component In namesstring[] Column names Out domainsstring[][] Domains for categorical columns Out cross_validation_modelsKey[] Cross-validation models (model ids) Out cross_validation_predictionsKey[] Cross-validation predictions, one per cv model (deprecated, use cross_validation_holdout_predictions_frame_id instead) Out cross_validation_holdout_predictions_frame_idKey Cross-validation holdout predictions (full out-of-sample predictions on training data) Out cross_validation_fold_assignment_frame_idKey Cross-validation fold assignment (each row is assigned to one holdout fold) Out model_categoryenum Category of the model (e.g., Binomial) Out model_summaryTwoDimTable Model summary Out scoring_historyTwoDimTable Scoring history Out training_metricsModelMetrics Training data model metrics Out validation_metricsModelMetrics Validation data model metrics Out cross_validation_metricsModelMetrics Cross-validation model metrics Out cross_validation_metrics_summaryTwoDimTable Cross-validation model metrics summary Out statusstring Job status Out start_timelong Start time in milliseconds Out end_timelong End time in milliseconds Out run_timelong Runtime in milliseconds Out helpMap Help information for output fields Out

## GLRMModelV3

 model_idKey Model key In/Out parametersGLRMParameters The build parameters for the model (e.g. K for KMeans). Out outputGLRMOutput The build output for the model (e.g. the cluster centers for KMeans). Out compatible_framesstring[] Compatible frames, if requested Out checksumlong Checksum for all the things that go into building the Model. Out algostring The algo name for this Model. Out algo_full_namestring The pretty algo name for this Model (e.g., Generalized Linear Model, rather than GLM). Out response_column_namestring The response column name for this Model (if applicable). Is null otherwise. Out data_frameKey The Model’s training frame key Out timestamplong Timestamp for when this model was completed Out

## GLRMParametersV3

 transformenum Transformation of training data In kint Rank of matrix approximation In lossenum Numeric loss function In multi_lossenum Categorical loss function In loss_by_colenum[] Loss function by column (override) In loss_by_col_idxint[] Loss function by column index (override) In periodint Length of period (only used with periodic loss function) In regularization_xenum Regularization function for X matrix In regularization_yenum Regularization function for Y matrix In gamma_xdouble Regularization weight on X matrix In gamma_ydouble Regularization weight on Y matrix In max_iterationsint Maximum number of iterations In max_updatesint Maximum number of updates, defaults to 2*max_iterations In init_step_sizedouble Initial step size In min_step_sizedouble Minimum step size In seedlong RNG seed for initialization In initenum Initialization mode In svd_methodenum Method for computing SVD during initialization (Caution: Power and Randomized are currently experimental and unstable) In user_yKey User-specified initial Y In user_xKey User-specified initial X In loading_namestring Frame key to save resulting X In expand_user_yboolean Expand categorical columns in user-specified initial Y In impute_originalboolean Reconstruct original training data by reversing transform In recover_svdboolean Recover singular values and eigenvectors of XY In distributionenum Distribution function In tweedie_powerdouble Tweedie power for Tweedie regression, must be between 1 and 2. In quantile_alphadouble Desired quantile for Quantile regression, must be between 0 and 1. In huber_alphadouble Desired quantile for Huber/M-regression (threshold between quadratic and linear loss, must be between 0 and 1). In model_idKey Destination id for this model; auto-generated if not specified. In/Out training_frameKey Id of the training data frame (Not required, to allow initial validation of model parameters). In/Out validation_frameKey Id of the validation data frame. In/Out nfoldsint Number of folds for N-fold cross-validation (0 to disable or >= 2). In/Out keep_cross_validation_predictionsboolean Whether to keep the predictions of the cross-validation models. In/Out keep_cross_validation_fold_assignmentboolean Whether to keep the cross-validation fold assignment. In/Out parallelize_cross_validationboolean Allow parallel training of cross-validation models In/Out response_columnVecSpecifier Response variable column. In/Out weights_columnVecSpecifier Column with observation weights. Giving some observation a weight of zero is equivalent to excluding it from the dataset; giving an observation a relative weight of 2 is equivalent to repeating that row twice. Negative weights are not allowed. In/Out offset_columnVecSpecifier Offset column. This will be added to the combination of columns before applying the link function. In/Out fold_columnVecSpecifier Column with cross-validation fold index assignment per observation. In/Out fold_assignmentenum Cross-validation fold assignment scheme, if fold_column is not specified. The ‘Stratified’ option will stratify the folds based on the response variable, for classification problems. In/Out categorical_encodingenum Encoding scheme for categorical features In/Out ignored_columnsstring[] Names of columns to ignore for training. In/Out ignore_const_colsboolean Ignore constant columns. In/Out score_each_iterationboolean Whether to score during each iteration of model training. In/Out checkpointKey Model checkpoint to resume training with. In/Out stopping_roundsint Early stopping based on convergence of stopping_metric. Stop if simple moving average of length k of the stopping_metric does not improve for k:=stopping_rounds scoring events (0 to disable) In/Out max_runtime_secsdouble Maximum allowed runtime in seconds for model training. Use 0 to disable. In/Out stopping_metricenum Metric to use for early stopping (AUTO: logloss for classification, deviance for regression) In/Out stopping_tolerancedouble Relative tolerance for metric-based stopping criterion (stop if relative improvement is not at least this much) In/Out

## GLRMV3

 parametersGLRMParameters Model builder parameters. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In algostring The algo name for this ModelBuilder. Out algo_full_namestring The pretty algo name for this ModelBuilder (e.g., Generalized Linear Model, rather than GLM). Out can_buildenum[] Model categories this ModelBuilder can build. Out supervisedboolean Indicator whether the model is supervised or not. Out visibilityenum Should the builder always be visible, be marked as beta, or only visible if the user starts up with the experimental flag? Out jobJob Job Key Out messagesValidationMessage[] Parameter validation messages Out error_countint Count of parameter validation errors Out __http_statusint HTTP status to return for this build. Out

(No fields)

## GramV3

 XKey source data In WVecSpecifier weight vector In use_all_factor_levelsboolean use all factor levels when doing 1-hot encoding In standardizeboolean standardize data In skip_missingboolean skip missing values In destination_frameKey Destination key for the resulting matrix. Out

## GrepModelOutputV3

 matchesstring[] Matching strings In offsetslong[] Byte offsets of matches In namesstring[] Column names Out domainsstring[][] Domains for categorical columns Out cross_validation_modelsKey[] Cross-validation models (model ids) Out cross_validation_predictionsKey[] Cross-validation predictions, one per cv model (deprecated, use cross_validation_holdout_predictions_frame_id instead) Out cross_validation_holdout_predictions_frame_idKey Cross-validation holdout predictions (full out-of-sample predictions on training data) Out cross_validation_fold_assignment_frame_idKey Cross-validation fold assignment (each row is assigned to one holdout fold) Out model_categoryenum Category of the model (e.g., Binomial) Out model_summaryTwoDimTable Model summary Out scoring_historyTwoDimTable Scoring history Out training_metricsModelMetrics Training data model metrics Out validation_metricsModelMetrics Validation data model metrics Out cross_validation_metricsModelMetrics Cross-validation model metrics Out cross_validation_metrics_summaryTwoDimTable Cross-validation model metrics summary Out statusstring Job status Out start_timelong Start time in milliseconds Out end_timelong End time in milliseconds Out run_timelong Runtime in milliseconds Out helpMap Help information for output fields Out

## GrepModelV3

 model_idKey Model key In/Out parametersGrepParameters The build parameters for the model (e.g. K for KMeans). Out outputGrepOutput The build output for the model (e.g. the cluster centers for KMeans). Out compatible_framesstring[] Compatible frames, if requested Out checksumlong Checksum for all the things that go into building the Model. Out algostring The algo name for this Model. Out algo_full_namestring The pretty algo name for this Model (e.g., Generalized Linear Model, rather than GLM). Out response_column_namestring The response column name for this Model (if applicable). Is null otherwise. Out data_frameKey The Model’s training frame key Out timestamplong Timestamp for when this model was completed Out

## GrepParametersV3

 regexstring regex In distributionenum Distribution function In tweedie_powerdouble Tweedie power for Tweedie regression, must be between 1 and 2. In quantile_alphadouble Desired quantile for Quantile regression, must be between 0 and 1. In huber_alphadouble Desired quantile for Huber/M-regression (threshold between quadratic and linear loss, must be between 0 and 1). In model_idKey Destination id for this model; auto-generated if not specified. In/Out training_frameKey Id of the training data frame (Not required, to allow initial validation of model parameters). In/Out validation_frameKey Id of the validation data frame. In/Out nfoldsint Number of folds for N-fold cross-validation (0 to disable or >= 2). In/Out keep_cross_validation_predictionsboolean Whether to keep the predictions of the cross-validation models. In/Out keep_cross_validation_fold_assignmentboolean Whether to keep the cross-validation fold assignment. In/Out parallelize_cross_validationboolean Allow parallel training of cross-validation models In/Out response_columnVecSpecifier Response variable column. In/Out weights_columnVecSpecifier Column with observation weights. Giving some observation a weight of zero is equivalent to excluding it from the dataset; giving an observation a relative weight of 2 is equivalent to repeating that row twice. Negative weights are not allowed. In/Out offset_columnVecSpecifier Offset column. This will be added to the combination of columns before applying the link function. In/Out fold_columnVecSpecifier Column with cross-validation fold index assignment per observation. In/Out fold_assignmentenum Cross-validation fold assignment scheme, if fold_column is not specified. The ‘Stratified’ option will stratify the folds based on the response variable, for classification problems. In/Out categorical_encodingenum Encoding scheme for categorical features In/Out ignored_columnsstring[] Names of columns to ignore for training. In/Out ignore_const_colsboolean Ignore constant columns. In/Out score_each_iterationboolean Whether to score during each iteration of model training. In/Out checkpointKey Model checkpoint to resume training with. In/Out stopping_roundsint Early stopping based on convergence of stopping_metric. Stop if simple moving average of length k of the stopping_metric does not improve for k:=stopping_rounds scoring events (0 to disable) In/Out max_runtime_secsdouble Maximum allowed runtime in seconds for model training. Use 0 to disable. In/Out stopping_metricenum Metric to use for early stopping (AUTO: logloss for classification, deviance for regression) In/Out stopping_tolerancedouble Relative tolerance for metric-based stopping criterion (stop if relative improvement is not at least this much) In/Out

## GrepV3

 parametersGrepParameters Model builder parameters. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In algostring The algo name for this ModelBuilder. Out algo_full_namestring The pretty algo name for this ModelBuilder (e.g., Generalized Linear Model, rather than GLM). Out can_buildenum[] Model categories this ModelBuilder can build. Out supervisedboolean Indicator whether the model is supervised or not. Out visibilityenum Should the builder always be visible, be marked as beta, or only visible if the user starts up with the experimental flag? Out jobJob Job Key Out messagesValidationMessage[] Parameter validation messages Out error_countint Count of parameter validation errors Out __http_statusint HTTP status to return for this build. Out

## GridKeyV3

 namestring Name (string representation) for this Key. In/Out typestring Name (string representation) for the type of Keyed this Key points to. In/Out URLstring URL for the resource that this Key points to, if one exists. In/Out

## GridSchemaV99

 grid_idKey Grid id In model_idsKey[] Model IDs built by a grid search In sort_bystring Model performance metric to sort by. Examples: logloss, residual_deviance, mse, rmse, mae,rmsle, auc, r2, f1, recall, precision, accuracy, mcc, err, err_count, lift_top_group, max_per_class_error In/Out decreasingboolean Specify whether sort order should be decreasing. In/Out hyper_namesstring[] Used hyper parameters. Out failed_paramsParameters[] List of failed parameters Out failure_detailsstring[] List of detailed failure messages Out failure_stack_tracesstring[] List of detailed failure stack traces Out failed_raw_paramsstring[][] List of raw parameters causing model building failure Out training_metricsModelMetrics[] Training model metrics for the returned models; only returned if sort_by is set Out validation_metricsModelMetrics[] Validation model metrics for the returned models; only returned if sort_by is set Out cross_validation_metricsModelMetrics[] Cross validation model metrics for the returned models; only returned if sort_by is set Out cross_validation_metrics_summaryTwoDimTable[] Cross validation model metrics summary for the returned models; only returned if sort_by is set Out summary_tableTwoDimTable Summary Out scoring_historyTwoDimTable Scoring history Out

## GridSearchSchema

 parametersParameters Basic model builder parameters. In hyper_parametersMap Grid search parameters. In/Out grid_idKey Destination id for this grid; auto-generated if not specified. In/Out search_criteriaHyperSpaceSearchCriteria Hyperparameter search criteria, including strategy and early stopping directives. If it is not given, exhaustive Cartesian is used. In/Out total_modelsint Number of all models generated by grid search. Out jobJob Job Key. Out

## GridsV99

 gridsGrid[] Grids Out

## H2OErrorV3

 timestamplong Milliseconds since the epoch for the time that this H2OError instance was created. Generally this is a short time since the underlying error ocurred. Out error_urlstring Error url Out msgstring Message intended for the end user (a data scientist). Out dev_msgstring Potentially more detailed message intended for a developer (e.g. a front end engineer or someone designing a language binding). Out http_statusint HTTP status code for this error. Out valuesMap Any values that are relevant to reporting or handling this error. Examples are a key name if the error is on a key, or a field name and object name if it’s on a specific field. Out exception_typestring Exception type, if any. Out exception_msgstring Raw exception message, if any. Out stacktracestring[] Stacktrace, if any. Out

## H2OModelBuilderErrorV3

 parametersParameters Model builder parameters. Out messagesValidationMessage[] Parameter validation messages Out error_countint Count of parameter validation errors Out timestamplong Milliseconds since the epoch for the time that this H2OError instance was created. Generally this is a short time since the underlying error ocurred. Out error_urlstring Error url Out msgstring Message intended for the end user (a data scientist). Out dev_msgstring Potentially more detailed message intended for a developer (e.g. a front end engineer or someone designing a language binding). Out http_statusint HTTP status code for this error. Out valuesMap Any values that are relevant to reporting or handling this error. Examples are a key name if the error is on a key, or a field name and object name if it’s on a specific field. Out exception_typestring Exception type, if any. Out exception_msgstring Raw exception message, if any. Out stacktracestring[] Stacktrace, if any. Out

## HeartBeatEvent

 sendsint number of sent heartbeats In recvsint number of received heartbeats In datestring Time when the event was recorded. Format is hh:mm:ss:ms In nanoslong Time in nanos In typeenum type of recorded event In

## HyperSpaceSearchCriteriaV99

 strategyenum Hyperparameter space search strategy. In/Out

## IOEvent

 io_flavorstring flavor of the recorded io (ice/hdfs/…) In nodestring node where this io event happened In datastring data info In datestring Time when the event was recorded. Format is hh:mm:ss:ms In nanoslong Time in nanos In typeenum type of recorded event In

## ImportFilesV3

 pathstring path In patternstring pattern In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In filesstring[] files Out destination_framesstring[] names Out failsstring[] fails Out delsstring[] dels Out

## ImportSQLTableV99

 connection_urlstring connection_url In tablestring table In select_querystring select_query In usernamestring username In passwordstring password In columnsstring columns In optimizeboolean optimize In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In

## InitIDV3

 _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In session_keystring Session ID In/Out

## InputSchemaV4

 _fieldsstring Filter on the set of output fields: if you set _fields=”foo,bar,baz”, then only those fields will be included in the output; or you can specify _fields=”-goo,gee” to include all fields except goo and gee. If the result contains nested data structures, then you can refer to the fields within those structures as well. For example if you specify _fields=”foo(oof),bar(-rab)”, then only fields foo and bar will be included, and within foo there will be only field oof, whereas within bar all fields except rab will be reported. In

## InteractionV3

 _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In destKey destination key In/Out source_frameKey Input data frame In/Out factor_columnsstring[] Factor columns In/Out pairwiseboolean Whether to create pairwise quadratic interactions between factors (otherwise create one higher-order interaction). Only applicable if there are 3 or more factors. In/Out max_factorsint Max. number of factor levels in pair-wise interaction terms (if enforced, one extra catch-all factor will be made) In/Out min_occurrenceint Min. occurrence threshold for factor levels in pair-wise interaction terms In/Out

## IoStatsEntry

 backendstring Back end type Out store_countlong Number of store events Out store_byteslong Cumulative stored bytes Out delete_countlong Number of delete events Out load_countlong Number of load events Out load_byteslong Cumulative loaded bytes Out

## JStackV3

 _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In tracesDStackTrace[] Stacktraces Out

## JobKeyV3

 namestring Name (string representation) for this Key. In/Out typestring Name (string representation) for the type of Keyed this Key points to. In/Out URLstring URL for the resource that this Key points to, if one exists. In/Out

## JobV3

 keyKey Job Key In descriptionstring Job description In destKey destination key In/Out statusstring job status Out progressfloat progress, from 0 to 1 Out progress_msgstring current progress status description Out start_timelong Start time Out mseclong Runtime in milliseconds Out warningsstring[] exception Out exceptionstring exception Out stacktracestring stacktrace Out ready_for_viewboolean ready for view Out

## JobsV3

 job_idKey Optional Job identifier In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In jobsJob[] jobs Out

## KMeansModelOutputV3

 centersTwoDimTable Cluster Centers[k][features] In centers_stdTwoDimTable Cluster Centers[k][features] on Standardized Data In namesstring[] Column names Out domainsstring[][] Domains for categorical columns Out cross_validation_modelsKey[] Cross-validation models (model ids) Out cross_validation_predictionsKey[] Cross-validation predictions, one per cv model (deprecated, use cross_validation_holdout_predictions_frame_id instead) Out cross_validation_holdout_predictions_frame_idKey Cross-validation holdout predictions (full out-of-sample predictions on training data) Out cross_validation_fold_assignment_frame_idKey Cross-validation fold assignment (each row is assigned to one holdout fold) Out model_categoryenum Category of the model (e.g., Binomial) Out model_summaryTwoDimTable Model summary Out scoring_historyTwoDimTable Scoring history Out training_metricsModelMetrics Training data model metrics Out validation_metricsModelMetrics Validation data model metrics Out cross_validation_metricsModelMetrics Cross-validation model metrics Out cross_validation_metrics_summaryTwoDimTable Cross-validation model metrics summary Out statusstring Job status Out start_timelong Start time in milliseconds Out end_timelong End time in milliseconds Out run_timelong Runtime in milliseconds Out helpMap Help information for output fields Out

## KMeansModelV3

 model_idKey Model key In/Out parametersKMeansParameters The build parameters for the model (e.g. K for KMeans). Out outputKMeansOutput The build output for the model (e.g. the cluster centers for KMeans). Out compatible_framesstring[] Compatible frames, if requested Out checksumlong Checksum for all the things that go into building the Model. Out algostring The algo name for this Model. Out algo_full_namestring The pretty algo name for this Model (e.g., Generalized Linear Model, rather than GLM). Out response_column_namestring The response column name for this Model (if applicable). Is null otherwise. Out data_frameKey The Model’s training frame key Out timestamplong Timestamp for when this model was completed Out

## KMeansParametersV3

 user_pointsKey User-specified points In max_iterationsint Maximum training iterations (if estimate_k is enabled, then this is for each inner Lloyds iteration) In standardizeboolean Standardize columns before computing distances In seedlong RNG Seed In initenum Initialization mode In estimate_kboolean Whether to estimate the number of clusters (<=k) iteratively and deterministically. In distributionenum Distribution function In tweedie_powerdouble Tweedie power for Tweedie regression, must be between 1 and 2. In quantile_alphadouble Desired quantile for Quantile regression, must be between 0 and 1. In huber_alphadouble Desired quantile for Huber/M-regression (threshold between quadratic and linear loss, must be between 0 and 1). In kint The max. number of clusters. If estimate_k is disabled, the model will find k centroids, otherwise it will find up to k centroids. In/Out model_idKey Destination id for this model; auto-generated if not specified. In/Out training_frameKey Id of the training data frame (Not required, to allow initial validation of model parameters). In/Out validation_frameKey Id of the validation data frame. In/Out nfoldsint Number of folds for N-fold cross-validation (0 to disable or >= 2). In/Out keep_cross_validation_predictionsboolean Whether to keep the predictions of the cross-validation models. In/Out keep_cross_validation_fold_assignmentboolean Whether to keep the cross-validation fold assignment. In/Out parallelize_cross_validationboolean Allow parallel training of cross-validation models In/Out response_columnVecSpecifier Response variable column. In/Out weights_columnVecSpecifier Column with observation weights. Giving some observation a weight of zero is equivalent to excluding it from the dataset; giving an observation a relative weight of 2 is equivalent to repeating that row twice. Negative weights are not allowed. In/Out offset_columnVecSpecifier Offset column. This will be added to the combination of columns before applying the link function. In/Out fold_columnVecSpecifier Column with cross-validation fold index assignment per observation. In/Out fold_assignmentenum Cross-validation fold assignment scheme, if fold_column is not specified. The ‘Stratified’ option will stratify the folds based on the response variable, for classification problems. In/Out categorical_encodingenum Encoding scheme for categorical features In/Out ignored_columnsstring[] Names of columns to ignore for training. In/Out ignore_const_colsboolean Ignore constant columns. In/Out score_each_iterationboolean Whether to score during each iteration of model training. In/Out checkpointKey Model checkpoint to resume training with. In/Out stopping_roundsint Early stopping based on convergence of stopping_metric. Stop if simple moving average of length k of the stopping_metric does not improve for k:=stopping_rounds scoring events (0 to disable) In/Out max_runtime_secsdouble Maximum allowed runtime in seconds for model training. Use 0 to disable. In/Out stopping_metricenum Metric to use for early stopping (AUTO: logloss for classification, deviance for regression) In/Out stopping_tolerancedouble Relative tolerance for metric-based stopping criterion (stop if relative improvement is not at least this much) In/Out

## KMeansV3

 parametersKMeansParameters Model builder parameters. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In algostring The algo name for this ModelBuilder. Out algo_full_namestring The pretty algo name for this ModelBuilder (e.g., Generalized Linear Model, rather than GLM). Out can_buildenum[] Model categories this ModelBuilder can build. Out supervisedboolean Indicator whether the model is supervised or not. Out visibilityenum Should the builder always be visible, be marked as beta, or only visible if the user starts up with the experimental flag? Out jobJob Job Key Out messagesValidationMessage[] Parameter validation messages Out error_countint Count of parameter validation errors Out __http_statusint HTTP status to return for this build. Out

## KeyV3

 namestring Name (string representation) for this Key. In/Out typestring Name (string representation) for the type of Keyed this Key points to. In/Out URLstring URL for the resource that this Key points to, if one exists. In/Out

## KillMinus3V3

 _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In

## ListRequestV4

 __schemastring Url describing the schema of the current object. In

## LogAndEchoV3

 messagestring Message to be Logged and Echoed In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In

## LogsV3

 nodeidxint Index of node to query ticks for (0-based). -1 means current node. In namestring Which specific log file to read from the log file directory. If left unspecified, the system chooses a default for you. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In logstring Content of log file Out

## MakeGLMModelV3

 modelKey source model In destKey destination key In namesstring[] coefficient names In betadouble[] new glm coefficients In thresholdfloat decision threshold for label-generation In

 numint Number for specifying an endpoint In http_methodstring HTTP method (GET, POST, DELETE) if fetching by path In pathstring Path for specifying an endpoint In classnamestring Class name, for fetching docs for a schema (DEPRECATED) In schemanamestring Schema name (e.g., DocsV1), for fetching docs for a schema In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In routesRoute[] List of endpoint routes Out schemasSchemaMetadata[] List of schemas Out markdownstring Table of Contents Markdown Out

## MissingInserterV3

 datasetKey dataset In fractiondouble Fraction of data to replace with a missing value In seedlong Seed In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In

## ModelBuilderSchema

 parametersParameters Model builder parameters. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In algostring The algo name for this ModelBuilder. Out algo_full_namestring The pretty algo name for this ModelBuilder (e.g., Generalized Linear Model, rather than GLM). Out can_buildenum[] Model categories this ModelBuilder can build. Out supervisedboolean Indicator whether the model is supervised or not. Out visibilityenum Should the builder always be visible, be marked as beta, or only visible if the user starts up with the experimental flag? Out jobJob Job Key Out messagesValidationMessage[] Parameter validation messages Out error_countint Count of parameter validation errors Out __http_statusint HTTP status to return for this build. Out

## ModelBuilderV3

 parametersParameters Model builder parameters. Out messagesValidationMessage[] Info, warning and error messages; NOTE: can be appended to while the Job is running Out error_countint Count of error messages Out

## ModelBuildersV3

 algostring Algo of ModelBuilder of interest In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In model_buildersMap ModelBuilders Out

## ModelExportV3

 model_idKey Name of Model of interest In dirstring Destination file (hdfs, s3, local) In forceboolean Overwrite destination file in case it exists or throw exception if set to false. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In

## ModelIdV3

 model_idstring Model ID Out

## ModelImportV3

 model_idKey Save imported model under given key into DKV. In dirstring Source directory (hdfs, s3, local) containing serialized model In forceboolean Override existing model in case it exists or throw exception if set to false In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In

## ModelInfoV4

 algostring Algorithm name, such as ‘gbm’, ‘deeplearning’, etc. In maturitystring Development status of the algorithm: alpha, beta, or stable. In have_pojoboolean Does the model support generation of POJOs? In have_mojoboolean Does the model support generation of MOJOs? In mojo_versionstring Mojo version number for this algorithm. In __schemastring Url describing the schema of the current object. In

## ModelKeyV3

 namestring Name (string representation) for this Key. In/Out typestring Name (string representation) for the type of Keyed this Key points to. In/Out URLstring URL for the resource that this Key points to, if one exists. In/Out

## ModelMetricsAutoEncoderV3

 nobslong Number of observations. In modelKey The model used for this scoring run. In/Out model_checksumlong The checksum for the model used for this scoring run. In/Out frameKey The frame used for this scoring run. In/Out frame_checksumlong The checksum for the frame used for this scoring run. In/Out descriptionstring Optional description for this scoring run (to note out-of-bag, sampled data, etc.) Out model_categoryenum The category (e.g., Clustering) for the model used for this scoring run. Out scoring_timelong The time in mS since the epoch for the start of this scoring run. Out predictionsFrame Predictions Frame. Out MSEdouble The Mean Squared Error of the prediction for this scoring run. Out RMSEdouble The Root Mean Squared Error of the prediction for this scoring run. Out

## ModelMetricsBaseV3

 nobslong Number of observations. In modelKey The model used for this scoring run. In/Out model_checksumlong The checksum for the model used for this scoring run. In/Out frameKey The frame used for this scoring run. In/Out frame_checksumlong The checksum for the frame used for this scoring run. In/Out descriptionstring Optional description for this scoring run (to note out-of-bag, sampled data, etc.) Out model_categoryenum The category (e.g., Clustering) for the model used for this scoring run. Out scoring_timelong The time in mS since the epoch for the start of this scoring run. Out predictionsFrame Predictions Frame. Out MSEdouble The Mean Squared Error of the prediction for this scoring run. Out RMSEdouble The Root Mean Squared Error of the prediction for this scoring run. Out

## ModelMetricsBinomialGLMV3

 nobslong Number of observations. In modelKey The model used for this scoring run. In/Out model_checksumlong The checksum for the model used for this scoring run. In/Out frameKey The frame used for this scoring run. In/Out frame_checksumlong The checksum for the frame used for this scoring run. In/Out residual_deviancedouble residual deviance Out null_deviancedouble null deviance Out AICdouble AIC Out null_degrees_of_freedomlong null DOF Out residual_degrees_of_freedomlong residual DOF Out r2double The R^2 for this scoring run. Out loglossdouble The logarithmic loss for this scoring run. Out AUCdouble The AUC for this scoring run. Out Ginidouble The Gini score for this scoring run. Out mean_per_class_errordouble The mean misclassification error per class. Out domainstring[] The class labels of the response. Out thresholds_and_metric_scoresTwoDimTable The Metrics for various thresholds. Out max_criteria_and_metric_scoresTwoDimTable The Metrics for various criteria. Out gains_lift_tableTwoDimTable Gains and Lift table. Out descriptionstring Optional description for this scoring run (to note out-of-bag, sampled data, etc.) Out model_categoryenum The category (e.g., Clustering) for the model used for this scoring run. Out scoring_timelong The time in mS since the epoch for the start of this scoring run. Out predictionsFrame Predictions Frame. Out MSEdouble The Mean Squared Error of the prediction for this scoring run. Out RMSEdouble The Root Mean Squared Error of the prediction for this scoring run. Out

## ModelMetricsBinomialV3

 nobslong Number of observations. In modelKey The model used for this scoring run. In/Out model_checksumlong The checksum for the model used for this scoring run. In/Out frameKey The frame used for this scoring run. In/Out frame_checksumlong The checksum for the frame used for this scoring run. In/Out r2double The R^2 for this scoring run. Out loglossdouble The logarithmic loss for this scoring run. Out AUCdouble The AUC for this scoring run. Out Ginidouble The Gini score for this scoring run. Out mean_per_class_errordouble The mean misclassification error per class. Out domainstring[] The class labels of the response. Out thresholds_and_metric_scoresTwoDimTable The Metrics for various thresholds. Out max_criteria_and_metric_scoresTwoDimTable The Metrics for various criteria. Out gains_lift_tableTwoDimTable Gains and Lift table. Out descriptionstring Optional description for this scoring run (to note out-of-bag, sampled data, etc.) Out model_categoryenum The category (e.g., Clustering) for the model used for this scoring run. Out scoring_timelong The time in mS since the epoch for the start of this scoring run. Out predictionsFrame Predictions Frame. Out MSEdouble The Mean Squared Error of the prediction for this scoring run. Out RMSEdouble The Root Mean Squared Error of the prediction for this scoring run. Out

## ModelMetricsClusteringV3

 tot_withinssdouble Within Cluster Sum of Square Error In totssdouble Total Sum of Square Error to Grand Mean In betweenssdouble Between Cluster Sum of Square Error In centroid_statsTwoDimTable Centroid Statistics In nobslong Number of observations. In modelKey The model used for this scoring run. In/Out model_checksumlong The checksum for the model used for this scoring run. In/Out frameKey The frame used for this scoring run. In/Out frame_checksumlong The checksum for the frame used for this scoring run. In/Out descriptionstring Optional description for this scoring run (to note out-of-bag, sampled data, etc.) Out model_categoryenum The category (e.g., Clustering) for the model used for this scoring run. Out scoring_timelong The time in mS since the epoch for the start of this scoring run. Out predictionsFrame Predictions Frame. Out MSEdouble The Mean Squared Error of the prediction for this scoring run. Out RMSEdouble The Root Mean Squared Error of the prediction for this scoring run. Out

## ModelMetricsGLRMV99

 numerrdouble Sum of Squared Error (Numeric Cols) In caterrdouble Misclassification Error (Categorical Cols) In numcntlong Number of Non-Missing Numeric Values In catcntlong Number of Non-Missing Categorical Values In nobslong Number of observations. In modelKey The model used for this scoring run. In/Out model_checksumlong The checksum for the model used for this scoring run. In/Out frameKey The frame used for this scoring run. In/Out frame_checksumlong The checksum for the frame used for this scoring run. In/Out descriptionstring Optional description for this scoring run (to note out-of-bag, sampled data, etc.) Out model_categoryenum The category (e.g., Clustering) for the model used for this scoring run. Out scoring_timelong The time in mS since the epoch for the start of this scoring run. Out predictionsFrame Predictions Frame. Out MSEdouble The Mean Squared Error of the prediction for this scoring run. Out RMSEdouble The Root Mean Squared Error of the prediction for this scoring run. Out

## ModelMetricsListSchemaV3

 modelKey Key of Model of interest (optional) In frameKey Key of Frame of interest (optional) In reconstruction_errorboolean Compute reconstruction error (optional, only for Deep Learning AutoEncoder models) In reconstruction_error_per_featureboolean Compute reconstruction error per feature (optional, only for Deep Learning AutoEncoder models) In deep_features_hidden_layerint Extract Deep Features for given hidden layer (optional, only for Deep Learning models) In reconstruct_trainboolean Reconstruct original training frame (optional, only for GLRM models) In project_archetypesboolean Project GLRM archetypes back into original feature space (optional, only for GLRM models) In reverse_transformboolean Reverse transformation applied during training to model output (optional, only for GLRM models) In leaf_node_assignmentboolean Return the leaf node assignment (optional, only for DRF/GBM models) In exemplar_indexint Retrieve all members for a given exemplar (optional, only for Aggregator models) In deviancesboolean Compute the deviances per row (optional, only for classification or regression models) In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In predictions_frameKey Key of predictions frame, if predictions are requested (optional) In/Out deviances_frameKey Key for the frame containing per-observation deviances (optional) In/Out model_metricsModelMetrics[] ModelMetrics Out

## ModelMetricsMakerSchemaV3

 predictions_framestring Predictions Frame. In/Out actuals_framestring Actuals Frame. In/Out domainstring[] Domain (for classification). In/Out distributionenum Distribution (for regression). In/Out model_metricsModelMetrics Model Metrics. Out

## ModelMetricsMultinomialGLMV3

 nobslong Number of observations. In modelKey The model used for this scoring run. In/Out model_checksumlong The checksum for the model used for this scoring run. In/Out frameKey The frame used for this scoring run. In/Out frame_checksumlong The checksum for the frame used for this scoring run. In/Out residual_deviancedouble residual deviance Out null_deviancedouble null deviance Out AICdouble AIC Out null_degrees_of_freedomlong null DOF Out residual_degrees_of_freedomlong residual DOF Out r2double The R^2 for this scoring run. Out hit_ratio_tableTwoDimTable The hit ratio table for this scoring run. Out cmConfusionMatrix The ConfusionMatrix object for this scoring run. Out loglossdouble The logarithmic loss for this scoring run. Out mean_per_class_errordouble The mean misclassification error per class. Out descriptionstring Optional description for this scoring run (to note out-of-bag, sampled data, etc.) Out model_categoryenum The category (e.g., Clustering) for the model used for this scoring run. Out scoring_timelong The time in mS since the epoch for the start of this scoring run. Out predictionsFrame Predictions Frame. Out MSEdouble The Mean Squared Error of the prediction for this scoring run. Out RMSEdouble The Root Mean Squared Error of the prediction for this scoring run. Out

## ModelMetricsMultinomialV3

 nobslong Number of observations. In modelKey The model used for this scoring run. In/Out model_checksumlong The checksum for the model used for this scoring run. In/Out frameKey The frame used for this scoring run. In/Out frame_checksumlong The checksum for the frame used for this scoring run. In/Out r2double The R^2 for this scoring run. Out hit_ratio_tableTwoDimTable The hit ratio table for this scoring run. Out cmConfusionMatrix The ConfusionMatrix object for this scoring run. Out loglossdouble The logarithmic loss for this scoring run. Out mean_per_class_errordouble The mean misclassification error per class. Out descriptionstring Optional description for this scoring run (to note out-of-bag, sampled data, etc.) Out model_categoryenum The category (e.g., Clustering) for the model used for this scoring run. Out scoring_timelong The time in mS since the epoch for the start of this scoring run. Out predictionsFrame Predictions Frame. Out MSEdouble The Mean Squared Error of the prediction for this scoring run. Out RMSEdouble The Root Mean Squared Error of the prediction for this scoring run. Out

## ModelMetricsPCAV3

 nobslong Number of observations. In modelKey The model used for this scoring run. In/Out model_checksumlong The checksum for the model used for this scoring run. In/Out frameKey The frame used for this scoring run. In/Out frame_checksumlong The checksum for the frame used for this scoring run. In/Out descriptionstring Optional description for this scoring run (to note out-of-bag, sampled data, etc.) Out model_categoryenum The category (e.g., Clustering) for the model used for this scoring run. Out scoring_timelong The time in mS since the epoch for the start of this scoring run. Out predictionsFrame Predictions Frame. Out MSEdouble The Mean Squared Error of the prediction for this scoring run. Out RMSEdouble The Root Mean Squared Error of the prediction for this scoring run. Out

## ModelMetricsRegressionGLMV3

 nobslong Number of observations. In modelKey The model used for this scoring run. In/Out model_checksumlong The checksum for the model used for this scoring run. In/Out frameKey The frame used for this scoring run. In/Out frame_checksumlong The checksum for the frame used for this scoring run. In/Out residual_deviancedouble residual deviance Out null_deviancedouble null deviance Out AICdouble AIC Out null_degrees_of_freedomlong null DOF Out residual_degrees_of_freedomlong residual DOF Out r2double The R^2 for this scoring run. Out mean_residual_deviancedouble The mean residual deviance for this scoring run. Out maedouble The mean absolute error for this scoring run. Out rmsledouble The root mean squared log error for this scoring run. Out descriptionstring Optional description for this scoring run (to note out-of-bag, sampled data, etc.) Out model_categoryenum The category (e.g., Clustering) for the model used for this scoring run. Out scoring_timelong The time in mS since the epoch for the start of this scoring run. Out predictionsFrame Predictions Frame. Out MSEdouble The Mean Squared Error of the prediction for this scoring run. Out RMSEdouble The Root Mean Squared Error of the prediction for this scoring run. Out

## ModelMetricsRegressionV3

 nobslong Number of observations. In modelKey The model used for this scoring run. In/Out model_checksumlong The checksum for the model used for this scoring run. In/Out frameKey The frame used for this scoring run. In/Out frame_checksumlong The checksum for the frame used for this scoring run. In/Out r2double The R^2 for this scoring run. Out mean_residual_deviancedouble The mean residual deviance for this scoring run. Out maedouble The mean absolute error for this scoring run. Out rmsledouble The root mean squared log error for this scoring run. Out descriptionstring Optional description for this scoring run (to note out-of-bag, sampled data, etc.) Out model_categoryenum The category (e.g., Clustering) for the model used for this scoring run. Out scoring_timelong The time in mS since the epoch for the start of this scoring run. Out predictionsFrame Predictions Frame. Out MSEdouble The Mean Squared Error of the prediction for this scoring run. Out RMSEdouble The Root Mean Squared Error of the prediction for this scoring run. Out

## ModelMetricsSVDV99

 nobslong Number of observations. In modelKey The model used for this scoring run. In/Out model_checksumlong The checksum for the model used for this scoring run. In/Out frameKey The frame used for this scoring run. In/Out frame_checksumlong The checksum for the frame used for this scoring run. In/Out descriptionstring Optional description for this scoring run (to note out-of-bag, sampled data, etc.) Out model_categoryenum The category (e.g., Clustering) for the model used for this scoring run. Out scoring_timelong The time in mS since the epoch for the start of this scoring run. Out predictionsFrame Predictions Frame. Out MSEdouble The Mean Squared Error of the prediction for this scoring run. Out RMSEdouble The Root Mean Squared Error of the prediction for this scoring run. Out

## ModelOutputSchemaV3

 namesstring[] Column names Out domainsstring[][] Domains for categorical columns Out cross_validation_modelsKey[] Cross-validation models (model ids) Out cross_validation_predictionsKey[] Cross-validation predictions, one per cv model (deprecated, use cross_validation_holdout_predictions_frame_id instead) Out cross_validation_holdout_predictions_frame_idKey Cross-validation holdout predictions (full out-of-sample predictions on training data) Out cross_validation_fold_assignment_frame_idKey Cross-validation fold assignment (each row is assigned to one holdout fold) Out model_categoryenum Category of the model (e.g., Binomial) Out model_summaryTwoDimTable Model summary Out scoring_historyTwoDimTable Scoring history Out training_metricsModelMetrics Training data model metrics Out validation_metricsModelMetrics Validation data model metrics Out cross_validation_metricsModelMetrics Cross-validation model metrics Out cross_validation_metrics_summaryTwoDimTable Cross-validation model metrics summary Out statusstring Job status Out start_timelong Start time in milliseconds Out end_timelong End time in milliseconds Out run_timelong Runtime in milliseconds Out helpMap Help information for output fields Out

## ModelParameterSchemaV3

 is_member_of_framesstring[] For Vec-type fields this is the set of other Vec-type fields which must contain mutually exclusive values; for example, for a SupervisedModel the response_column must be mutually exclusive with the weights_column In is_mutually_exclusive_withstring[] For Vec-type fields this is the set of Frame-type fields which must contain the named column; for example, for a SupervisedModel the response_column must be in both the training_frame and (if it’s set) the validation_frame In namestring name in the JSON, e.g. “lambda” Out labelstring [DEPRECATED] same as name. Out helpstring help for the UI, e.g. “regularization multiplier, typically used for foo bar baz etc.” Out requiredboolean the field is required Out typestring Java type, e.g. “double” Out default_valuePolymorphic default value, e.g. 1 Out actual_valuePolymorphic actual value as set by the user and / or modified by the ModelBuilder, e.g., 10 Out levelstring the importance of the parameter, used by the UI, e.g. “critical”, “extended” or “expert” Out valuesstring[] list of valid values for use by the front-end Out gridableboolean Parameter can be used in grid call Out

 distributionenum Distribution function In tweedie_powerdouble Tweedie power for Tweedie regression, must be between 1 and 2. In quantile_alphadouble Desired quantile for Quantile regression, must be between 0 and 1. In huber_alphadouble Desired quantile for Huber/M-regression (threshold between quadratic and linear loss, must be between 0 and 1). In model_idKey Destination id for this model; auto-generated if not specified. In/Out training_frameKey Id of the training data frame (Not required, to allow initial validation of model parameters). In/Out validation_frameKey Id of the validation data frame. In/Out nfoldsint Number of folds for N-fold cross-validation (0 to disable or >= 2). In/Out keep_cross_validation_predictionsboolean Whether to keep the predictions of the cross-validation models. In/Out keep_cross_validation_fold_assignmentboolean Whether to keep the cross-validation fold assignment. In/Out parallelize_cross_validationboolean Allow parallel training of cross-validation models In/Out response_columnVecSpecifier Response variable column. In/Out weights_columnVecSpecifier Column with observation weights. Giving some observation a weight of zero is equivalent to excluding it from the dataset; giving an observation a relative weight of 2 is equivalent to repeating that row twice. Negative weights are not allowed. In/Out offset_columnVecSpecifier Offset column. This will be added to the combination of columns before applying the link function. In/Out fold_columnVecSpecifier Column with cross-validation fold index assignment per observation. In/Out fold_assignmentenum Cross-validation fold assignment scheme, if fold_column is not specified. The ‘Stratified’ option will stratify the folds based on the response variable, for classification problems. In/Out categorical_encodingenum Encoding scheme for categorical features In/Out ignored_columnsstring[] Names of columns to ignore for training. In/Out ignore_const_colsboolean Ignore constant columns. In/Out score_each_iterationboolean Whether to score during each iteration of model training. In/Out checkpointKey Model checkpoint to resume training with. In/Out stopping_roundsint Early stopping based on convergence of stopping_metric. Stop if simple moving average of length k of the stopping_metric does not improve for k:=stopping_rounds scoring events (0 to disable) In/Out max_runtime_secsdouble Maximum allowed runtime in seconds for model training. Use 0 to disable. In/Out stopping_metricenum Metric to use for early stopping (AUTO: logloss for classification, deviance for regression) In/Out stopping_tolerancedouble Relative tolerance for metric-based stopping criterion (stop if relative improvement is not at least this much) In/Out

## ModelSchemaBaseV3

 model_idKey Model key In/Out algostring The algo name for this Model. Out algo_full_namestring The pretty algo name for this Model (e.g., Generalized Linear Model, rather than GLM). Out response_column_namestring The response column name for this Model (if applicable). Is null otherwise. Out data_frameKey The Model’s training frame key Out timestamplong Timestamp for when this model was completed Out

## ModelSchemaV3

 model_idKey Model key In/Out parametersParameters The build parameters for the model (e.g. K for KMeans). Out outputOutput The build output for the model (e.g. the cluster centers for KMeans). Out compatible_framesstring[] Compatible frames, if requested Out checksumlong Checksum for all the things that go into building the Model. Out algostring The algo name for this Model. Out algo_full_namestring The pretty algo name for this Model (e.g., Generalized Linear Model, rather than GLM). Out response_column_namestring The response column name for this Model (if applicable). Is null otherwise. Out data_frameKey The Model’s training frame key Out timestamplong Timestamp for when this model was completed Out

## ModelSynopsisV3

 model_idKey Model key In/Out algostring The algo name for this Model. Out algo_full_namestring The pretty algo name for this Model (e.g., Generalized Linear Model, rather than GLM). Out response_column_namestring The response column name for this Model (if applicable). Is null otherwise. Out data_frameKey The Model’s training frame key Out timestamplong Timestamp for when this model was completed Out

## ModelsInfoV4

 modelsModelBuilder[] Generic information about each model supported in H2O. In __schemastring Url describing the schema of the current object. In

## ModelsV3

 model_idKey Name of Model of interest In previewboolean Return potentially abridged model suitable for viewing in a browser In find_compatible_framesboolean Find and return compatible frames? In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In modelsModel[] Models Out compatible_framesFrame[] Compatible frames Out

## NaiveBayesModelOutputV3

 levelsstring[] Categorical levels of the response In aprioriTwoDimTable A-priori probabilities of the response In pcondTwoDimTable[] Conditional probabilities of the predictors In namesstring[] Column names Out domainsstring[][] Domains for categorical columns Out cross_validation_modelsKey[] Cross-validation models (model ids) Out cross_validation_predictionsKey[] Cross-validation predictions, one per cv model (deprecated, use cross_validation_holdout_predictions_frame_id instead) Out cross_validation_holdout_predictions_frame_idKey Cross-validation holdout predictions (full out-of-sample predictions on training data) Out cross_validation_fold_assignment_frame_idKey Cross-validation fold assignment (each row is assigned to one holdout fold) Out model_categoryenum Category of the model (e.g., Binomial) Out model_summaryTwoDimTable Model summary Out scoring_historyTwoDimTable Scoring history Out training_metricsModelMetrics Training data model metrics Out validation_metricsModelMetrics Validation data model metrics Out cross_validation_metricsModelMetrics Cross-validation model metrics Out cross_validation_metrics_summaryTwoDimTable Cross-validation model metrics summary Out statusstring Job status Out start_timelong Start time in milliseconds Out end_timelong End time in milliseconds Out run_timelong Runtime in milliseconds Out helpMap Help information for output fields Out

## NaiveBayesModelV3

 model_idKey Model key In/Out parametersNaiveBayesParameters The build parameters for the model (e.g. K for KMeans). Out outputNaiveBayesOutput The build output for the model (e.g. the cluster centers for KMeans). Out compatible_framesstring[] Compatible frames, if requested Out checksumlong Checksum for all the things that go into building the Model. Out algostring The algo name for this Model. Out algo_full_namestring The pretty algo name for this Model (e.g., Generalized Linear Model, rather than GLM). Out response_column_namestring The response column name for this Model (if applicable). Is null otherwise. Out data_frameKey The Model’s training frame key Out timestamplong Timestamp for when this model was completed Out

## NaiveBayesParametersV3

 laplacedouble Laplace smoothing parameter In min_sdevdouble Min. standard deviation to use for observations with not enough data In eps_sdevdouble Cutoff below which standard deviation is replaced with min_sdev In min_probdouble Min. probability to use for observations with not enough data In eps_probdouble Cutoff below which probability is replaced with min_prob In compute_metricsboolean Compute metrics on training data In distributionenum Distribution function In tweedie_powerdouble Tweedie power for Tweedie regression, must be between 1 and 2. In quantile_alphadouble Desired quantile for Quantile regression, must be between 0 and 1. In huber_alphadouble Desired quantile for Huber/M-regression (threshold between quadratic and linear loss, must be between 0 and 1). In balance_classesboolean Balance training data class counts via over/under-sampling (for imbalanced data). In/Out class_sampling_factorsfloat[] Desired over/under-sampling ratios per class (in lexicographic order). If not specified, sampling factors will be automatically computed to obtain class balance during training. Requires balance_classes. In/Out max_after_balance_sizefloat Maximum relative size of the training data after balancing class counts (can be less than 1.0). Requires balance_classes. In/Out max_confusion_matrix_sizeint [Deprecated] Maximum size (# classes) for confusion matrices to be printed in the Logs In/Out max_hit_ratio_kint Max. number (top K) of predictions to use for hit ratio computation (for multi-class only, 0 to disable) In/Out seedlong Seed for pseudo random number generator (only used for cross-validation and fold_assignment=”Random” or “AUTO”) In/Out model_idKey Destination id for this model; auto-generated if not specified. In/Out training_frameKey Id of the training data frame (Not required, to allow initial validation of model parameters). In/Out validation_frameKey Id of the validation data frame. In/Out nfoldsint Number of folds for N-fold cross-validation (0 to disable or >= 2). In/Out keep_cross_validation_predictionsboolean Whether to keep the predictions of the cross-validation models. In/Out keep_cross_validation_fold_assignmentboolean Whether to keep the cross-validation fold assignment. In/Out parallelize_cross_validationboolean Allow parallel training of cross-validation models In/Out response_columnVecSpecifier Response variable column. In/Out weights_columnVecSpecifier Column with observation weights. Giving some observation a weight of zero is equivalent to excluding it from the dataset; giving an observation a relative weight of 2 is equivalent to repeating that row twice. Negative weights are not allowed. In/Out offset_columnVecSpecifier Offset column. This will be added to the combination of columns before applying the link function. In/Out fold_columnVecSpecifier Column with cross-validation fold index assignment per observation. In/Out fold_assignmentenum Cross-validation fold assignment scheme, if fold_column is not specified. The ‘Stratified’ option will stratify the folds based on the response variable, for classification problems. In/Out categorical_encodingenum Encoding scheme for categorical features In/Out ignored_columnsstring[] Names of columns to ignore for training. In/Out ignore_const_colsboolean Ignore constant columns. In/Out score_each_iterationboolean Whether to score during each iteration of model training. In/Out checkpointKey Model checkpoint to resume training with. In/Out stopping_roundsint Early stopping based on convergence of stopping_metric. Stop if simple moving average of length k of the stopping_metric does not improve for k:=stopping_rounds scoring events (0 to disable) In/Out max_runtime_secsdouble Maximum allowed runtime in seconds for model training. Use 0 to disable. In/Out stopping_metricenum Metric to use for early stopping (AUTO: logloss for classification, deviance for regression) In/Out stopping_tolerancedouble Relative tolerance for metric-based stopping criterion (stop if relative improvement is not at least this much) In/Out

## NaiveBayesV3

 parametersNaiveBayesParameters Model builder parameters. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In algostring The algo name for this ModelBuilder. Out algo_full_namestring The pretty algo name for this ModelBuilder (e.g., Generalized Linear Model, rather than GLM). Out can_buildenum[] Model categories this ModelBuilder can build. Out supervisedboolean Indicator whether the model is supervised or not. Out visibilityenum Should the builder always be visible, be marked as beta, or only visible if the user starts up with the experimental flag? Out jobJob Job Key Out messagesValidationMessage[] Parameter validation messages Out error_countint Count of parameter validation errors Out __http_statusint HTTP status to return for this build. Out

## NetworkBenchV3

 _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In resultsTwoDimTable[] NetworkBenchResults Out

## NetworkEvent

 is_sendboolean Boolean flag distinguishing between sends (true) and receives(false) In protocolstring network protocol (UDP/TCP) In msg_typestring UDP type (exec,ack, ackack,… In fromstring Sending node In tostring Receiving node In datastring Pretty print of the first few bytes of the msg payload. Contains class name for tasks. In datestring Time when the event was recorded. Format is hh:mm:ss:ms In nanoslong Time in nanos In typeenum type of recorded event In

## NetworkTestV3

 _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In microseconds_collectivedouble[] Collective broadcast/reduce times in microseconds (for each message size) Out bandwidths_collectivedouble[] Collective bandwidths in Bytes/sec (for each message size, for each node) Out microsecondsdouble[][] Round-trip times in microseconds (for each message size, for each node) Out bandwidthsdouble[][] Bi-directional bandwidths in Bytes/sec (for each message size, for each node) Out nodesstring[] Nodes Out tableTwoDimTable NetworkTestResults Out

## NodePersistentStorageEntryV3

 categorystring Category name Out namestring Key name Out sizelong Size in bytes of value Out timestamp_millislong Epoch time in milliseconds of when the value was written Out

## NodePersistentStorageV3

 _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In categorystring Category name In/Out namestring Key name In/Out valuestring Value In/Out configuredboolean Configured Out existsboolean Exists Out entriesIced[] List of entries Out

## NodeV3

 h2ostring IP Out ip_portstring IP address and port in the form a.b.c.d:e Out healthyboolean (now-last_ping)local keys< Out free_disklong Free disk Out max_disklong Max disk Out rpcs_activeint Active Remote Procedure Calls Out fjthrdsshort[] F/J Thread count, by priority Out fjqueueshort[] F/J Task count, by priority Out tcps_activeint Open TCP connections Out open_fdsint Open File Descripters Out gflopsdouble Linpack GFlops Out mem_bwdouble Memory Bandwidth Out

## OutputSchemaV4

 __schemastring Url describing the schema of the current object. In

## PCAModelOutputV3

 importanceTwoDimTable Standard deviation and importance of each principal component In eigenvectorsTwoDimTable Principal components matrix In objectivedouble Final value of GLRM squared loss function In namesstring[] Column names Out domainsstring[][] Domains for categorical columns Out cross_validation_modelsKey[] Cross-validation models (model ids) Out cross_validation_predictionsKey[] Cross-validation predictions, one per cv model (deprecated, use cross_validation_holdout_predictions_frame_id instead) Out cross_validation_holdout_predictions_frame_idKey Cross-validation holdout predictions (full out-of-sample predictions on training data) Out cross_validation_fold_assignment_frame_idKey Cross-validation fold assignment (each row is assigned to one holdout fold) Out model_categoryenum Category of the model (e.g., Binomial) Out model_summaryTwoDimTable Model summary Out scoring_historyTwoDimTable Scoring history Out training_metricsModelMetrics Training data model metrics Out validation_metricsModelMetrics Validation data model metrics Out cross_validation_metricsModelMetrics Cross-validation model metrics Out cross_validation_metrics_summaryTwoDimTable Cross-validation model metrics summary Out statusstring Job status Out start_timelong Start time in milliseconds Out end_timelong End time in milliseconds Out run_timelong Runtime in milliseconds Out helpMap Help information for output fields Out

## PCAModelV3

 model_idKey Model key In/Out parametersPCAParameters The build parameters for the model (e.g. K for KMeans). Out outputPCAOutput The build output for the model (e.g. the cluster centers for KMeans). Out compatible_framesstring[] Compatible frames, if requested Out checksumlong Checksum for all the things that go into building the Model. Out algostring The algo name for this Model. Out algo_full_namestring The pretty algo name for this Model (e.g., Generalized Linear Model, rather than GLM). Out response_column_namestring The response column name for this Model (if applicable). Is null otherwise. Out data_frameKey The Model’s training frame key Out timestamplong Timestamp for when this model was completed Out

## PCAParametersV3

 transformenum Transformation of training data In pca_methodenum Method for computing PCA (Caution: Power and GLRM are currently experimental and unstable) In distributionenum Distribution function In tweedie_powerdouble Tweedie power for Tweedie regression, must be between 1 and 2. In quantile_alphadouble Desired quantile for Quantile regression, must be between 0 and 1. In huber_alphadouble Desired quantile for Huber/M-regression (threshold between quadratic and linear loss, must be between 0 and 1). In kint Rank of matrix approximation In/Out max_iterationsint Maximum training iterations In/Out seedlong RNG seed for initialization In/Out use_all_factor_levelsboolean Whether first factor level is included in each categorical expansion In/Out compute_metricsboolean Whether to compute metrics on the training data In/Out impute_missingboolean Whether to impute missing entries with the column mean In/Out model_idKey Destination id for this model; auto-generated if not specified. In/Out training_frameKey Id of the training data frame (Not required, to allow initial validation of model parameters). In/Out validation_frameKey Id of the validation data frame. In/Out nfoldsint Number of folds for N-fold cross-validation (0 to disable or >= 2). In/Out keep_cross_validation_predictionsboolean Whether to keep the predictions of the cross-validation models. In/Out keep_cross_validation_fold_assignmentboolean Whether to keep the cross-validation fold assignment. In/Out parallelize_cross_validationboolean Allow parallel training of cross-validation models In/Out response_columnVecSpecifier Response variable column. In/Out weights_columnVecSpecifier Column with observation weights. Giving some observation a weight of zero is equivalent to excluding it from the dataset; giving an observation a relative weight of 2 is equivalent to repeating that row twice. Negative weights are not allowed. In/Out offset_columnVecSpecifier Offset column. This will be added to the combination of columns before applying the link function. In/Out fold_columnVecSpecifier Column with cross-validation fold index assignment per observation. In/Out fold_assignmentenum Cross-validation fold assignment scheme, if fold_column is not specified. The ‘Stratified’ option will stratify the folds based on the response variable, for classification problems. In/Out categorical_encodingenum Encoding scheme for categorical features In/Out ignored_columnsstring[] Names of columns to ignore for training. In/Out ignore_const_colsboolean Ignore constant columns. In/Out score_each_iterationboolean Whether to score during each iteration of model training. In/Out checkpointKey Model checkpoint to resume training with. In/Out stopping_roundsint Early stopping based on convergence of stopping_metric. Stop if simple moving average of length k of the stopping_metric does not improve for k:=stopping_rounds scoring events (0 to disable) In/Out max_runtime_secsdouble Maximum allowed runtime in seconds for model training. Use 0 to disable. In/Out stopping_metricenum Metric to use for early stopping (AUTO: logloss for classification, deviance for regression) In/Out stopping_tolerancedouble Relative tolerance for metric-based stopping criterion (stop if relative improvement is not at least this much) In/Out

## PCAV3

 parametersPCAParameters Model builder parameters. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In algostring The algo name for this ModelBuilder. Out algo_full_namestring The pretty algo name for this ModelBuilder (e.g., Generalized Linear Model, rather than GLM). Out can_buildenum[] Model categories this ModelBuilder can build. Out supervisedboolean Indicator whether the model is supervised or not. Out visibilityenum Should the builder always be visible, be marked as beta, or only visible if the user starts up with the experimental flag? Out jobJob Job Key Out messagesValidationMessage[] Parameter validation messages Out error_countint Count of parameter validation errors Out __http_statusint HTTP status to return for this build. Out

## ParseSVMLightV3

 destination_frameKey Final frame name In source_framesKey[] Source frames In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In

## ParseSetupV3

 _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In source_framesKey[] Source frames In/Out parse_typeenum Parser type In/Out separatorbyte Field separator In/Out single_quotesboolean Single quotes In/Out check_headerint Check header: 0 means guess, +1 means 1st line is header not data, -1 means 1st line is data not header In/Out column_namesstring[] Column names In/Out column_typesstring[] Value types for columns In/Out na_stringsstring[][] NA strings for columns In/Out column_name_filterstring Regex for names of columns to return In/Out column_offsetint Column offset to return In/Out column_countint Number of columns to return In/Out total_filtered_column_countint Total number of columns we would return with no column pagination In/Out destination_framestring Suggested name Out header_lineslong Number of header lines found Out number_columnsint Number of columns Out datastring[][] Sample data Out warningsstring[] Warnings Out chunk_sizeint Size of individual parse tasks Out

## ParseV3

 destination_frameKey Final frame name In source_framesKey[] Source frames In parse_typeenum Parser type In separatorbyte Field separator In single_quotesboolean Single Quotes In check_headerint Check header: 0 means guess, +1 means 1st line is header not data, -1 means 1st line is data not header In number_columnsint Number of columns In column_namesstring[] Column names In column_typesstring[] Value types for columns In domainsstring[][] Domains for categorical columns In na_stringsstring[][] NA strings for columns In chunk_sizeint Size of individual parse tasks In delete_on_doneboolean Delete input key after parse In blockingboolean Block until the parse completes (as opposed to returning early and requiring polling In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In jobJob Parse job Out rowslong Rows Out

## PartialDependenceKeyV3

 namestring Name (string representation) for this Key. In/Out typestring Name (string representation) for the type of Keyed this Key points to. In/Out URLstring URL for the resource that this Key points to, if one exists. In/Out

## PartialDependenceV3

 destination_keyKey Key to store the destination In model_idKey Model In/Out frame_idKey Frame In/Out colsstring[] Column(s) In/Out nbinsint Number of bins In/Out partial_dependence_dataTwoDimTable[] Partial Dependence Data Out

## ProfilerNodeEntryV3

 stacktracestring Stack trace Out countint Profile Count Out

## ProfilerNodeV3

 node_namestring Node names Out timestamplong Timestamp (millis since epoch) Out entriesIced[] Profile entry list Out

## ProfilerV3

 depthint Stack trace depth In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In nodesIced[] (No description available) Out

## QuantileParametersV3

 probsdouble[] Probabilities for quantiles In combine_methodenum How to combine quantiles for even sample sizes In distributionenum Distribution function In tweedie_powerdouble Tweedie power for Tweedie regression, must be between 1 and 2. In quantile_alphadouble Desired quantile for Quantile regression, must be between 0 and 1. In huber_alphadouble Desired quantile for Huber/M-regression (threshold between quadratic and linear loss, must be between 0 and 1). In model_idKey Destination id for this model; auto-generated if not specified. In/Out training_frameKey Id of the training data frame (Not required, to allow initial validation of model parameters). In/Out validation_frameKey Id of the validation data frame. In/Out nfoldsint Number of folds for N-fold cross-validation (0 to disable or >= 2). In/Out keep_cross_validation_predictionsboolean Whether to keep the predictions of the cross-validation models. In/Out keep_cross_validation_fold_assignmentboolean Whether to keep the cross-validation fold assignment. In/Out parallelize_cross_validationboolean Allow parallel training of cross-validation models In/Out response_columnVecSpecifier Response variable column. In/Out weights_columnVecSpecifier Column with observation weights. Giving some observation a weight of zero is equivalent to excluding it from the dataset; giving an observation a relative weight of 2 is equivalent to repeating that row twice. Negative weights are not allowed. In/Out offset_columnVecSpecifier Offset column. This will be added to the combination of columns before applying the link function. In/Out fold_columnVecSpecifier Column with cross-validation fold index assignment per observation. In/Out fold_assignmentenum Cross-validation fold assignment scheme, if fold_column is not specified. The ‘Stratified’ option will stratify the folds based on the response variable, for classification problems. In/Out categorical_encodingenum Encoding scheme for categorical features In/Out ignored_columnsstring[] Names of columns to ignore for training. In/Out ignore_const_colsboolean Ignore constant columns. In/Out score_each_iterationboolean Whether to score during each iteration of model training. In/Out checkpointKey Model checkpoint to resume training with. In/Out stopping_roundsint Early stopping based on convergence of stopping_metric. Stop if simple moving average of length k of the stopping_metric does not improve for k:=stopping_rounds scoring events (0 to disable) In/Out max_runtime_secsdouble Maximum allowed runtime in seconds for model training. Use 0 to disable. In/Out stopping_metricenum Metric to use for early stopping (AUTO: logloss for classification, deviance for regression) In/Out stopping_tolerancedouble Relative tolerance for metric-based stopping criterion (stop if relative improvement is not at least this much) In/Out

## QuantileV3

 parametersQuantileParameters Model builder parameters. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In algostring The algo name for this ModelBuilder. Out algo_full_namestring The pretty algo name for this ModelBuilder (e.g., Generalized Linear Model, rather than GLM). Out can_buildenum[] Model categories this ModelBuilder can build. Out supervisedboolean Indicator whether the model is supervised or not. Out visibilityenum Should the builder always be visible, be marked as beta, or only visible if the user starts up with the experimental flag? Out jobJob Job Key Out messagesValidationMessage[] Parameter validation messages Out error_countint Count of parameter validation errors Out __http_statusint HTTP status to return for this build. Out

## RandomDiscreteValueSearchCriteriaV99

 seedlong Seed for random number generator; set to a value other than -1 for reproducibility. In/Out max_modelsint Maximum number of models to build (optional). In/Out max_runtime_secsdouble Maximum time to spend building models (optional). In/Out stopping_roundsint Early stopping based on convergence of stopping_metric. Stop if simple moving average of length k of the stopping_metric does not improve for k:=stopping_rounds scoring events (0 to disable) In/Out stopping_metricenum Metric to use for early stopping (AUTO: logloss for classification, deviance for regression) In/Out stopping_tolerancedouble Relative tolerance for metric-based stopping criterion Relative tolerance for metric-based stopping criterion (stop if relative improvement is not at least this much) In/Out strategyenum Hyperparameter space search strategy. In/Out

## RapidsExpressionV3

 namestring (Class) name of the language construct In is_abstractboolean If true, then this is not a standalone construct but purely a grouping level. In patternstring Code fragment pattern. In descriptionstring Description of the functionality provided by this language construct. In subIced[] List of language constructs that grouped under this one. In

## RapidsFrameV3

 aststring A Rapids AstRoot expression In session_idstring Session key In idstring [DEPRECATED] Key name to assign Frame results In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In keyKey Frame result Out num_rowslong Rows in Frame result Out num_colsint Columns in Frame result Out

## RapidsFunctionV3

 aststring A Rapids AstRoot expression In session_idstring Session key In idstring [DEPRECATED] Key name to assign Frame results In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In funstrstring Function result Out

## RapidsHelpV3

 syntaxIced Rapids language, organized in a form of a tree (so that similar constructs are grouped together. Out

## RapidsNumberV3

 aststring A Rapids AstRoot expression In session_idstring Session key In idstring [DEPRECATED] Key name to assign Frame results In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In scalardouble Number result Out

## RapidsNumbersV3

 aststring A Rapids AstRoot expression In session_idstring Session key In idstring [DEPRECATED] Key name to assign Frame results In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In scalardouble[] Number array result Out

## RapidsSchemaV3

 aststring A Rapids AstRoot expression In session_idstring Session key In idstring [DEPRECATED] Key name to assign Frame results In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In

## RapidsStringV3

 aststring A Rapids AstRoot expression In session_idstring Session key In idstring [DEPRECATED] Key name to assign Frame results In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In stringstring String result Out

## RapidsStringsV3

 aststring A Rapids AstRoot expression In session_idstring Session key In idstring [DEPRECATED] Key name to assign Frame results In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In stringstring[] String array result Out

## RapidsV99

 aststring An Abstract Syntax Tree. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In errorstring Parsing error, if any Out scalardouble Scalar result Out funstrstring Function result Out stringstring String result Out keyKey Result key Out num_rowslong Rows in Frame result Out num_colsint Columns in Frame result Out

## RemoveAllV3

 _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In

## RemoveV3

 keyKey Object to be removed. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In

## RequestSchemaV3

 _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In

## RouteV3

 http_methodstring (No description available) Out url_patternstring (No description available) Out summarystring (No description available) Out api_namestring (No description available) Out handler_classstring (No description available) Out handler_methodstring (No description available) Out input_schemastring (No description available) Out output_schemastring (No description available) Out path_paramsstring[] (No description available) Out markdownstring (No description available) Out

## SVDModelOutputV99

 v_keyKey Frame key of right singular vectors In ddouble[] Singular values In u_keyKey Frame key of left singular vectors In namesstring[] Column names Out domainsstring[][] Domains for categorical columns Out cross_validation_modelsKey[] Cross-validation models (model ids) Out cross_validation_predictionsKey[] Cross-validation predictions, one per cv model (deprecated, use cross_validation_holdout_predictions_frame_id instead) Out cross_validation_holdout_predictions_frame_idKey Cross-validation holdout predictions (full out-of-sample predictions on training data) Out cross_validation_fold_assignment_frame_idKey Cross-validation fold assignment (each row is assigned to one holdout fold) Out model_categoryenum Category of the model (e.g., Binomial) Out model_summaryTwoDimTable Model summary Out scoring_historyTwoDimTable Scoring history Out training_metricsModelMetrics Training data model metrics Out validation_metricsModelMetrics Validation data model metrics Out cross_validation_metricsModelMetrics Cross-validation model metrics Out cross_validation_metrics_summaryTwoDimTable Cross-validation model metrics summary Out statusstring Job status Out start_timelong Start time in milliseconds Out end_timelong End time in milliseconds Out run_timelong Runtime in milliseconds Out helpMap Help information for output fields Out

## SVDModelV99

 model_idKey Model key In/Out parametersSVDParameters The build parameters for the model (e.g. K for KMeans). Out outputSVDOutput The build output for the model (e.g. the cluster centers for KMeans). Out compatible_framesstring[] Compatible frames, if requested Out checksumlong Checksum for all the things that go into building the Model. Out algostring The algo name for this Model. Out algo_full_namestring The pretty algo name for this Model (e.g., Generalized Linear Model, rather than GLM). Out response_column_namestring The response column name for this Model (if applicable). Is null otherwise. Out data_frameKey The Model’s training frame key Out timestamplong Timestamp for when this model was completed Out

## SVDParametersV99

 transformenum Transformation of training data In svd_methodenum Method for computing SVD (Caution: Power and Randomized are currently experimental and unstable) In nvint Number of right singular vectors In max_iterationsint Maximum iterations In seedlong RNG seed for k-means++ initialization In keep_uboolean Save left singular vectors? In u_namestring Frame key to save left singular vectors In distributionenum Distribution function In tweedie_powerdouble Tweedie power for Tweedie regression, must be between 1 and 2. In quantile_alphadouble Desired quantile for Quantile regression, must be between 0 and 1. In huber_alphadouble Desired quantile for Huber/M-regression (threshold between quadratic and linear loss, must be between 0 and 1). In use_all_factor_levelsboolean Whether first factor level is included in each categorical expansion In/Out model_idKey Destination id for this model; auto-generated if not specified. In/Out training_frameKey Id of the training data frame (Not required, to allow initial validation of model parameters). In/Out validation_frameKey Id of the validation data frame. In/Out nfoldsint Number of folds for N-fold cross-validation (0 to disable or >= 2). In/Out keep_cross_validation_predictionsboolean Whether to keep the predictions of the cross-validation models. In/Out keep_cross_validation_fold_assignmentboolean Whether to keep the cross-validation fold assignment. In/Out parallelize_cross_validationboolean Allow parallel training of cross-validation models In/Out response_columnVecSpecifier Response variable column. In/Out weights_columnVecSpecifier Column with observation weights. Giving some observation a weight of zero is equivalent to excluding it from the dataset; giving an observation a relative weight of 2 is equivalent to repeating that row twice. Negative weights are not allowed. In/Out offset_columnVecSpecifier Offset column. This will be added to the combination of columns before applying the link function. In/Out fold_columnVecSpecifier Column with cross-validation fold index assignment per observation. In/Out fold_assignmentenum Cross-validation fold assignment scheme, if fold_column is not specified. The ‘Stratified’ option will stratify the folds based on the response variable, for classification problems. In/Out categorical_encodingenum Encoding scheme for categorical features In/Out ignored_columnsstring[] Names of columns to ignore for training. In/Out ignore_const_colsboolean Ignore constant columns. In/Out score_each_iterationboolean Whether to score during each iteration of model training. In/Out checkpointKey Model checkpoint to resume training with. In/Out stopping_roundsint Early stopping based on convergence of stopping_metric. Stop if simple moving average of length k of the stopping_metric does not improve for k:=stopping_rounds scoring events (0 to disable) In/Out max_runtime_secsdouble Maximum allowed runtime in seconds for model training. Use 0 to disable. In/Out stopping_metricenum Metric to use for early stopping (AUTO: logloss for classification, deviance for regression) In/Out stopping_tolerancedouble Relative tolerance for metric-based stopping criterion (stop if relative improvement is not at least this much) In/Out

## SVDV99

 parametersSVDParameters Model builder parameters. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In algostring The algo name for this ModelBuilder. Out algo_full_namestring The pretty algo name for this ModelBuilder (e.g., Generalized Linear Model, rather than GLM). Out can_buildenum[] Model categories this ModelBuilder can build. Out supervisedboolean Indicator whether the model is supervised or not. Out visibilityenum Should the builder always be visible, be marked as beta, or only visible if the user starts up with the experimental flag? Out jobJob Job Key Out messagesValidationMessage[] Parameter validation messages Out error_countint Count of parameter validation errors Out __http_statusint HTTP status to return for this build. Out

 versionint Version number of the Schema. In namestring Simple name of the Schema. NOTE: the schema_names form a single namespace. In superclassstring Simple name of the superclass of the Schema. NOTE: the schema_names form a single namespace. In typestring Simple name of H2O type that this Schema represents. Must not be changed after creation (treat as final). In labelstring [DEPRECATED] This field is always the same as name. Out fieldsFieldMetadata[] All the public fields of the schema Out markdownstring Documentation for the schema in Markdown format with GitHub extensions Out

(No fields)

## SessionIdV4

 session_keystring Session ID In __schemastring Url describing the schema of the current object. In

## SharedTreeModelOutputV3

 variable_importancesTwoDimTable Variable Importances Out init_fdouble The Intercept term, the initial model function value to which trees make adjustments Out namesstring[] Column names Out domainsstring[][] Domains for categorical columns Out cross_validation_modelsKey[] Cross-validation models (model ids) Out cross_validation_predictionsKey[] Cross-validation predictions, one per cv model (deprecated, use cross_validation_holdout_predictions_frame_id instead) Out cross_validation_holdout_predictions_frame_idKey Cross-validation holdout predictions (full out-of-sample predictions on training data) Out cross_validation_fold_assignment_frame_idKey Cross-validation fold assignment (each row is assigned to one holdout fold) Out model_categoryenum Category of the model (e.g., Binomial) Out model_summaryTwoDimTable Model summary Out scoring_historyTwoDimTable Scoring history Out training_metricsModelMetrics Training data model metrics Out validation_metricsModelMetrics Validation data model metrics Out cross_validation_metricsModelMetrics Cross-validation model metrics Out cross_validation_metrics_summaryTwoDimTable Cross-validation model metrics summary Out statusstring Job status Out start_timelong Start time in milliseconds Out end_timelong End time in milliseconds Out run_timelong Runtime in milliseconds Out helpMap Help information for output fields Out

## SharedTreeModelV3

 model_idKey Model key In/Out parametersParameters The build parameters for the model (e.g. K for KMeans). Out outputOutput The build output for the model (e.g. the cluster centers for KMeans). Out compatible_framesstring[] Compatible frames, if requested Out checksumlong Checksum for all the things that go into building the Model. Out algostring The algo name for this Model. Out algo_full_namestring The pretty algo name for this Model (e.g., Generalized Linear Model, rather than GLM). Out response_column_namestring The response column name for this Model (if applicable). Is null otherwise. Out data_frameKey The Model’s training frame key Out timestamplong Timestamp for when this model was completed Out

## SharedTreeParametersV3

 ntreesint Number of trees. In max_depthint Maximum tree depth. In min_rowsdouble Fewest allowed (weighted) observations in a leaf. In nbinsint For numerical columns (real/int), build a histogram of (at least) this many bins, then split at the best point In nbins_top_levelint For numerical columns (real/int), build a histogram of (at most) this many bins at the root level, then decrease by factor of two per level In nbins_catsint For categorical columns (factors), build a histogram of this many bins, then split at the best point. Higher values can lead to more overfitting. In r2_stoppingdouble r2_stopping is no longer supported and will be ignored if set - please use stopping_rounds, stopping_metric and stopping_tolerance instead. Previous version of H2O would stop making trees when the R^2 metric equals or exceeds this In seedlong Seed for pseudo random number generator (if applicable) In build_tree_one_nodeboolean Run on one node only; no network overhead but fewer cpus used. Suitable for small datasets. In sample_ratedouble Row sample rate per tree (from 0.0 to 1.0) In sample_rate_per_classdouble[] Row sample rate per tree per class (from 0.0 to 1.0) In col_sample_rate_per_treedouble Column sample rate per tree (from 0.0 to 1.0) In col_sample_rate_change_per_leveldouble Relative change of the column sampling rate for every level (from 0.0 to 2.0) In score_tree_intervalint Score the model after every so many trees. Disabled if set to 0. In min_split_improvementdouble Minimum relative improvement in squared error reduction for a split to happen In histogram_typeenum What type of histogram to use for finding optimal split points In distributionenum Distribution function In tweedie_powerdouble Tweedie power for Tweedie regression, must be between 1 and 2. In quantile_alphadouble Desired quantile for Quantile regression, must be between 0 and 1. In huber_alphadouble Desired quantile for Huber/M-regression (threshold between quadratic and linear loss, must be between 0 and 1). In balance_classesboolean Balance training data class counts via over/under-sampling (for imbalanced data). In/Out class_sampling_factorsfloat[] Desired over/under-sampling ratios per class (in lexicographic order). If not specified, sampling factors will be automatically computed to obtain class balance during training. Requires balance_classes. In/Out max_after_balance_sizefloat Maximum relative size of the training data after balancing class counts (can be less than 1.0). Requires balance_classes. In/Out max_confusion_matrix_sizeint [Deprecated] Maximum size (# classes) for confusion matrices to be printed in the Logs In/Out max_hit_ratio_kint Max. number (top K) of predictions to use for hit ratio computation (for multi-class only, 0 to disable) In/Out model_idKey Destination id for this model; auto-generated if not specified. In/Out training_frameKey Id of the training data frame (Not required, to allow initial validation of model parameters). In/Out validation_frameKey Id of the validation data frame. In/Out nfoldsint Number of folds for N-fold cross-validation (0 to disable or >= 2). In/Out keep_cross_validation_predictionsboolean Whether to keep the predictions of the cross-validation models. In/Out keep_cross_validation_fold_assignmentboolean Whether to keep the cross-validation fold assignment. In/Out parallelize_cross_validationboolean Allow parallel training of cross-validation models In/Out response_columnVecSpecifier Response variable column. In/Out weights_columnVecSpecifier Column with observation weights. Giving some observation a weight of zero is equivalent to excluding it from the dataset; giving an observation a relative weight of 2 is equivalent to repeating that row twice. Negative weights are not allowed. In/Out offset_columnVecSpecifier Offset column. This will be added to the combination of columns before applying the link function. In/Out fold_columnVecSpecifier Column with cross-validation fold index assignment per observation. In/Out fold_assignmentenum Cross-validation fold assignment scheme, if fold_column is not specified. The ‘Stratified’ option will stratify the folds based on the response variable, for classification problems. In/Out categorical_encodingenum Encoding scheme for categorical features In/Out ignored_columnsstring[] Names of columns to ignore for training. In/Out ignore_const_colsboolean Ignore constant columns. In/Out score_each_iterationboolean Whether to score during each iteration of model training. In/Out checkpointKey Model checkpoint to resume training with. In/Out stopping_roundsint Early stopping based on convergence of stopping_metric. Stop if simple moving average of length k of the stopping_metric does not improve for k:=stopping_rounds scoring events (0 to disable) In/Out max_runtime_secsdouble Maximum allowed runtime in seconds for model training. Use 0 to disable. In/Out stopping_metricenum Metric to use for early stopping (AUTO: logloss for classification, deviance for regression) In/Out stopping_tolerancedouble Relative tolerance for metric-based stopping criterion (stop if relative improvement is not at least this much) In/Out

## SharedTreeV3

 parametersParameters Model builder parameters. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In algostring The algo name for this ModelBuilder. Out algo_full_namestring The pretty algo name for this ModelBuilder (e.g., Generalized Linear Model, rather than GLM). Out can_buildenum[] Model categories this ModelBuilder can build. Out supervisedboolean Indicator whether the model is supervised or not. Out visibilityenum Should the builder always be visible, be marked as beta, or only visible if the user starts up with the experimental flag? Out jobJob Job Key Out messagesValidationMessage[] Parameter validation messages Out error_countint Count of parameter validation errors Out __http_statusint HTTP status to return for this build. Out

## ShutdownV3

 _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In

## SplitFrameV3

 keyKey Job Key In datasetKey Dataset In ratiosdouble[] Split ratios - resulting number of split is ratios.length+1 In destination_framesKey[] Destination keys for each output frame split. In/Out

(No fields)

## TabulateV3

 datasetKey Dataset In nbins_predictorint Number of bins for predictor column In nbins_responseint Number of bins for response column In predictorVecSpecifier Predictor In/Out responseVecSpecifier Response In/Out weightVecSpecifier Observation weights (optional) In/Out count_tableTwoDimTable Counts table Out response_tableTwoDimTable Response table Out

## TimelineV3

 _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In nowlong Current time in millis. Out selfstring This node Out eventsIced[] recorded timeline events Out

## TreeStatsV3

 min_depthint minDepth In max_depthint maxDepth In mean_depthfloat meanDepth In min_leavesint minLeaves In max_leavesint maxLeaves In mean_leavesfloat meanLeaves In

## TwoDimTableV3

 namestring Table Name Out descriptionstring Table Description Out columnsIced[] Column Specification Out rowcountint Number of Rows Out dataPolymorphic[][] Table Data (col-major) Out

 srcstring training_frame In limitint limit In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In matchesstring[] matches Out

## UnlockKeysV3

 _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In

## ValidationMessageV3

 message_typestring Type of validation message (ERROR, WARN, INFO, HIDE) Out field_namestring Field to which the message applies Out messagestring Message text Out

## VarImpV3

 varimpfloat[] Variable importance of individual variables Out namesstring[] Names of variables Out

## WaterMeterCpuTicksV3

 nodeidxint Index of node to query ticks for (0-based) In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In cpu_tickslong[][] array of tick counts per core Out

## WaterMeterIoV3

 nodeidxint Index of node to query ticks for (0-based) In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In persist_statsIced[] array of IO info Out

## Word2VecModelOutputV3

 namesstring[] Column names Out domainsstring[][] Domains for categorical columns Out cross_validation_modelsKey[] Cross-validation models (model ids) Out cross_validation_predictionsKey[] Cross-validation predictions, one per cv model (deprecated, use cross_validation_holdout_predictions_frame_id instead) Out cross_validation_holdout_predictions_frame_idKey Cross-validation holdout predictions (full out-of-sample predictions on training data) Out cross_validation_fold_assignment_frame_idKey Cross-validation fold assignment (each row is assigned to one holdout fold) Out model_categoryenum Category of the model (e.g., Binomial) Out model_summaryTwoDimTable Model summary Out scoring_historyTwoDimTable Scoring history Out training_metricsModelMetrics Training data model metrics Out validation_metricsModelMetrics Validation data model metrics Out cross_validation_metricsModelMetrics Cross-validation model metrics Out cross_validation_metrics_summaryTwoDimTable Cross-validation model metrics summary Out statusstring Job status Out start_timelong Start time in milliseconds Out end_timelong End time in milliseconds Out run_timelong Runtime in milliseconds Out helpMap Help information for output fields Out

## Word2VecModelV3

 model_idKey Model key In/Out parametersWord2VecParameters The build parameters for the model (e.g. K for KMeans). Out outputWord2VecOutput The build output for the model (e.g. the cluster centers for KMeans). Out compatible_framesstring[] Compatible frames, if requested Out checksumlong Checksum for all the things that go into building the Model. Out algostring The algo name for this Model. Out algo_full_namestring The pretty algo name for this Model (e.g., Generalized Linear Model, rather than GLM). Out response_column_namestring The response column name for this Model (if applicable). Is null otherwise. Out data_frameKey The Model’s training frame key Out timestamplong Timestamp for when this model was completed Out

## Word2VecParametersV3

 vec_sizeint Set size of word vectors In window_sizeint Set max skip length between words In sent_sample_ratefloat Set threshold for occurrence of words. Those that appear with higher frequency in the training data will be randomly down-sampled; useful range is (0, 1e-5) In norm_modelenum Use Hierarchical Softmax In epochsint Number of training iterations to run In min_word_freqint This will discard words that appear less than times In init_learning_ratefloat Set the starting learning rate In word_modelenum Use the Skip-Gram model In distributionenum Distribution function In tweedie_powerdouble Tweedie power for Tweedie regression, must be between 1 and 2. In quantile_alphadouble Desired quantile for Quantile regression, must be between 0 and 1. In huber_alphadouble Desired quantile for Huber/M-regression (threshold between quadratic and linear loss, must be between 0 and 1). In model_idKey Destination id for this model; auto-generated if not specified. In/Out training_frameKey Id of the training data frame (Not required, to allow initial validation of model parameters). In/Out validation_frameKey Id of the validation data frame. In/Out nfoldsint Number of folds for N-fold cross-validation (0 to disable or >= 2). In/Out keep_cross_validation_predictionsboolean Whether to keep the predictions of the cross-validation models. In/Out keep_cross_validation_fold_assignmentboolean Whether to keep the cross-validation fold assignment. In/Out parallelize_cross_validationboolean Allow parallel training of cross-validation models In/Out response_columnVecSpecifier Response variable column. In/Out weights_columnVecSpecifier Column with observation weights. Giving some observation a weight of zero is equivalent to excluding it from the dataset; giving an observation a relative weight of 2 is equivalent to repeating that row twice. Negative weights are not allowed. In/Out offset_columnVecSpecifier Offset column. This will be added to the combination of columns before applying the link function. In/Out fold_columnVecSpecifier Column with cross-validation fold index assignment per observation. In/Out fold_assignmentenum Cross-validation fold assignment scheme, if fold_column is not specified. The ‘Stratified’ option will stratify the folds based on the response variable, for classification problems. In/Out categorical_encodingenum Encoding scheme for categorical features In/Out ignored_columnsstring[] Names of columns to ignore for training. In/Out ignore_const_colsboolean Ignore constant columns. In/Out score_each_iterationboolean Whether to score during each iteration of model training. In/Out checkpointKey Model checkpoint to resume training with. In/Out stopping_roundsint Early stopping based on convergence of stopping_metric. Stop if simple moving average of length k of the stopping_metric does not improve for k:=stopping_rounds scoring events (0 to disable) In/Out max_runtime_secsdouble Maximum allowed runtime in seconds for model training. Use 0 to disable. In/Out stopping_metricenum Metric to use for early stopping (AUTO: logloss for classification, deviance for regression) In/Out stopping_tolerancedouble Relative tolerance for metric-based stopping criterion (stop if relative improvement is not at least this much) In/Out

## Word2VecSynonymsV3

 modelKey Source word2vec Model In wordstring Target word to find synonyms for In countint Number of synonyms In synonymsstring[] Synonymous words In scoresdouble[] Similarity scores In

## Word2VecTransformV3

 modelKey Source word2vec Model In words_frameKey Words Frame In vectors_frameKey Word Vectors Frame Out

## Word2VecV3

 parametersWord2VecParameters Model builder parameters. In _exclude_fieldsstring Comma-separated list of JSON field paths to exclude from the result, used like: “/3/Frames?_exclude_fields=frames/frame_id/URL,__meta” In algostring The algo name for this ModelBuilder. Out algo_full_namestring The pretty algo name for this ModelBuilder (e.g., Generalized Linear Model, rather than GLM). Out can_buildenum[] Model categories this ModelBuilder can build. Out supervisedboolean Indicator whether the model is supervised or not. Out visibilityenum Should the builder always be visible, be marked as beta, or only visible if the user starts up with the experimental flag? Out jobJob Job Key Out messagesValidationMessage[] Parameter validation messages Out error_countint Count of parameter validation errors Out __http_statusint HTTP status to return for this build. Out