Hands-On: Automatic & Explainable Machine Learning with H2O

Abstract

H2O is an open source, distributed machine learning platform, designed to scale to very large datasets. H2O AutoML provides an easy-to-use interface (available in Python, R, Java and Scala) which automates data pre-processing, training and tuning a large selection of candidate models from a single function. The result of the AutoML run is a "leaderboard" of H2O models, which can be then be explained automatically using the newly released H2O Explainability interface.

The models can be explained individually, and also compared as a group, with global (model-wise) and local (row-wise) explanations are available. The explain function generates a list of explanations – individual units of explanation such as a Partial Dependence plot or a Variable Importance plot. Most of the explanations are visual – these plots can also be created by individual utility functions outside of the primary explain function. The visualization engine used in the R interface is ggplot2, and in Python, we use matplotlib.