This workflow explains how to train a GBM classifier in H2O, predict classes of new data and evaluate the performance. 1. Prepare: Load the IRIS data, import the resulting KNIME Table to H2O and partition the data for test and train set 30/70. 2. Learn: We learn the GBM Model using the H2O Gradient Boosting Machine Learner (Classification). We want H2O to build 1000 Trees using a multinominal distribution of the reponse, for it is a multilabel problem. All other model parameters are H2Os defaults. 3. Predict: Make predictions on new data using your model(s). In order to compute the Scoring metrics, we need to enable the "append individual class probabilities" parameter in the "H2O Predictor (Classification)" Node 4. Score: In order to evaluate our model, we asess the Classifiers accuracy by scoring the predictions made on the test data.
This is a companion discussion topic for the original entry at https://kni.me/w/lH10H9751NPJUUUn