H2O is an Open Source, Distributed, Fast & Scalable Machine Learning Platform: Deep Learning, Gradient Boosting (GBM) & XGBoost, Random Forest, Generalized Linear Modeling (GLM with Elastic Net), K-Means, PCA, Generalized Additive Models (GAM), RuleFit, Support Vector Machine (SVM), Stacked Ensembles, Automatic Machine Learning (AutoML), etc.
Add support of other distributions via the family argument. Since the H2O base learners support different potential families, we must auto-map any non-supported regression distributions to "gaussian". For example, if user specified "huber", that would be mapped to "gaussian" for all the H2O algos aside from DL (which explicitly supports "huber" distributions). We should allow full range (union) of family and distribution types.
Consider changing the family argument in h2o.ensemble() to distribution. GLM uses family and GBM and DL uses distribution. RF does not have an equivalent argument and instead infers distribution / ML task from the response column type.
family
argument. Since the H2O base learners support different potential families, we must auto-map any non-supported regression distributions to "gaussian". For example, if user specified "huber", that would be mapped to "gaussian" for all the H2O algos aside from DL (which explicitly supports "huber" distributions). We should allow full range (union) of family and distribution types.family
argument inh2o.ensemble()
todistribution
. GLM usesfamily
and GBM and DL usesdistribution
. RF does not have an equivalent argument and instead infers distribution / ML task from the response column type.