rf_paramGrid = ParamGridBuilder().addGrid(rf.maxDepth, [1, 5, 10]).addGrid(
rf.impurity, ['entropy', 'gini']).addGrid(rf.maxBins, [2, 10, 20]).build()
AUC_evaluator = BinaryClassificationEvaluator(rawPredictionCol='rawPrediction',
labelCol='label',
metricName="areaUnderROC")
accuracy_evaluator = MulticlassClassificationEvaluator(
labelCol="label", predictionCol="prediction", metricName="accuracy")
rf_paramGrid = CrossValidator(estimator=rf_pipeline,
estimatorParamMaps=rf_paramGrid,
evaluator=accuracy_evaluator,
numFolds=3)
cvModel_rf = rf_paramGrid.fit(trainingData)
BestPipeline_rf = cvModel_rf.bestModel
paramDict = {
param[0].name: param[1]
for param in BestPipeline_rf.stages[-1].extractParamMap().items()
}
# Here, we're converting the dictionary to a JSON object to make it easy to print. You can print it however you'd like
print(json.dumps(paramDict, indent=4))
predction_rfc = cvModel_rf.transform(testData)
AUC_rfc = AUC_evaluator.evaluate(predction_rfc)
