H2O scoring model performance metrics

In this example we take a look at the KNIME Nodes for H2O Scoring. There are different H2O Scorer Nodes in KNIME for different Machine Learning problems: The "H2O Regression Scorer" Node for regression problems and the Classification Scorer Nodes for Binominal and Multinominal classifiers. Depending on the problem (the type of the response) different scoring metrics are computed. 1. Classification scoring: In order to Score binominal or multinominal classifiers, we have to add columns with the per-class probabilities. This can be achieved by activating the setting "append individual class probabliities" in the H2O Predictor Nodes. The "H2O Binominal Scorer" Node computes classification metrics for response variables with two classes, resulting in Accuracy Statistics (Output Port 0), the confusion matrix (Output Port 1) and the gains lift (Output Port 2). The H2O Multinominal Scorer Node computes classification metrics (Output Port 0) for multilabel response variables (3 or more classes) as well as the confusion matrix (Output Port 1). 2. Regression scoring: The H2O Regression Scorer Node computes regression metrics like the RSME and R-Squared (Output Port 0).

This is a companion discussion topic for the original entry at https://kni.me/w/Pic0B9uiw-RB0zdT