H2O is an open source platform for building machine learning and predictive analytics models. The algorithms are written on H2O's distributed map-reduce framework. With H2O, the data is distributed across nodes, read in parallel, and stored in the memory in a compressed manner. This makes H2O extremely fast.

H2O's stacked ensemble method is an ensemble machine learning algorithm for supervised problems that finds the optimal combination of a collection of predictive algorithms using stacking. H2O's stacked ensemble supports regression, binary classification, and multiclass classification.

In this example, we'll take a look at how to use H2O's stacked ensemble to build a stacking model. We'll use the bank marketing dataset which is available in the Github.