The DecisionForestPerSegment function uses a training data set to create a predictive model for each partition of a partitioned input file. You can input the model to the DecisionForestPredictPerSegment (ML Engine) function, which uses it to make predictions.