Predict feature contributions - SHAP values on an H2O Model (only DRF, GBM and XGBoost models).

Returned H2OFrame has shape (#rows, #features + 1) - there is a feature contribution column for each input feature, the last column is the model bias (same value for each row). The sum of the feature contributions and the bias term is equal to the raw prediction of the model. Raw prediction of tree-based model is the sum of the predictions of the individual trees before the inverse link function is applied to get the actual prediction. For Gaussian distribution the sum of the contributions is equal to the model prediction.

predict_contributions.H2OModel(
  object,
  newdata,
  output_format = c("original", "compact"),
  ...
)

h2o.predict_contributions(
  object,
  newdata,
  output_format = c("original", "compact"),
  ...
)

Arguments

object	a fitted H2OModel object for which prediction is desired
newdata	An H2OFrame object in which to look for variables with which to predict.
output_format	Specify how to output feature contributions in XGBoost - XGBoost by default outputs contributions for 1-hot encoded features, specifying a compact output format will produce a per-feature contribution. Defaults to original.
...	additional arguments to pass on.

Value

Returns an H2OFrame contain feature contributions for each input row.

Details

Note: Multinomial classification models are currently not supported.

See also

h2o.gbm and h2o.randomForest for model generation in h2o.

Examples

# NOT RUN {
library(h2o)
h2o.init()
prostate_path <- system.file("extdata", "prostate.csv", package = "h2o")
prostate <- h2o.uploadFile(path = prostate_path)
prostate_gbm <- h2o.gbm(3:9, "AGE", prostate)
h2o.predict(prostate_gbm, prostate)
h2o.predict_contributions(prostate_gbm, prostate)
# }