Train RuleFit Model in Sparkling Water

RuleFit algorithm combines tree ensembles and linear models to take advantage of both methods: - the accuracy of a tree ensemble - the interpretability of a linear model

The general algorithm fits a tree ensemble to the data, builds a rule ensemble by traversing each tree, evaluates the rules on the data to build a rule feature set, and fits a sparse linear model (LASSO) to the rule feature set joined with the original feature set.

Sparkling Water provides API for H2O RuleFit in Scala and Python. The following sections describe how to train the RuleFit model in Sparkling Water in both languages. See also Parameters of H2ORuleFit and Details of H2ORuleFitMOJOModel.

Scala

First, let’s start Sparkling Shell as

./bin/sparkling-shell

Start H2O cluster inside the Spark environment

import ai.h2o.sparkling._
import java.net.URI
val hc = H2OContext.getOrCreate()

Parse the data using H2O and convert them to Spark Frame

import org.apache.spark.SparkFiles
spark.sparkContext.addFile("https://s3.amazonaws.com/h2o-public-test-data/smalldata/gbm_test/titanic.csv")
val rawSparkDF = spark.read
    .option("header", "true")
    .option("inferSchema", "true")
    .csv(SparkFiles.get("titanic.csv"))
val Array(trainingDF, testingDF) = rawSparkDF.randomSplit(Array(0.8, 0.2), seed = 1L)

Train the model. You can configure all the available RuleFit arguments using provided setters, such as the label column.

import ai.h2o.sparkling.ml.algos.H2ORuleFit
val estimator = new H2ORuleFit()
    .setMaxRuleLength(10)
    .setMaxNumRules(100)
    .setSeed(1)
    .setLabelCol("survived")
    .setColumnsToCategorical(Array("pclass", "survived"))
    .setFeaturesCols(Array("age", "sibsp", "parch", "fare", "sex", "pclass"))
val model = estimator.fit(trainingDF)

By default, the H2ORuleFit algorithm distinguishes between a classification and regression problem based on the type of the label column of the training dataset. If the label column is a string column, a classification model will be trained. If the label column is a numeric column, a regression model will be trained. If you don’t want be worried about column data types, you can explicitly identify the problem by using ai.h2o.sparkling.ml.algos.classification.H2OH2ORuleFitClassifier or ai.h2o.sparkling.ml.algos.regression.H2OH2ORuleFitRegressor instead.

Get trained rules

model.getRuleImportance().show(truncate = False)

You can also get model details via calling methods listed in Details of H2ORuleFitMOJOModel.

Run Predictions

model.transform(testingDF).show(false)

Python

First, let’s start PySparkling Shell as

./bin/pysparkling

Start H2O cluster inside the Spark environment

from pysparkling import *
hc = H2OContext.getOrCreate()

Parse the data using H2O and convert them to Spark Frame

import h2o
frame = h2o.import_file("https://s3.amazonaws.com/h2o-public-test-data/smalldata/gbm_test/titanic.csv")
sparkDF = hc.asSparkFrame(frame)
[trainingDF, testingDF] = sparkDF.randomSplit([0.8, 0.2], seed = 1)

Train the model. You can configure all the available RuleFit arguments using provided setters or constructor parameters, such as the label column.

from pysparkling.ml import H2ORuleFit
estimator = H2ORuleFit() \
    .setMaxRuleLength(10) \
    .setMaxNumRules(100) \
    .setSeed(1) \
    .setLabelCol("survived") \
    .setColumnsToCategorical(["pclass", "survived"]) \
    .setFeaturesCols(["age", "sibsp", "parch", "fare", "sex", "pclass"])
model = estimator.fit(trainingDF)

By default, the H2ORuleFit algorithm distinguishes between a classification and regression problem based on the type of the label column of the training dataset. If the label column is a string column, a classification model will be trained. If the label column is a numeric column, a regression model will be trained. If you don’t want to be worried about column data types, you can explicitly identify the problem by using H2ORuleFitClassifier or H2ORuleFitRegressor instead.

Get trained rules

model.getRuleImportance().show(truncate = False)

You can also get model details via calling methods listed in Details of H2ORuleFitMOJOModel.

Run Predictions

model.transform(testingDF).show(truncate = False)