Add an option to use logodds returned by base models to build a stacked ensemble

h2oai / h2o-3

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.

Apache License 2.0

6.92k stars 2k forks source link

Currently, the h2o.stackedensemble() function uses the predicted probabilities returned by base models as inputs to build the meta learner; however, when both the base models and meta learner are logistic regression models, this method eliminates the linear relationship between the original variables used by the base models and the final log odds returned by the meta learner.

Request if we could add an option to the function enabling user to determine whether to use the predicted probabilities or log odds to build the meta learner.

Thank you.

h2oai / h2o-3

Add an option to use logodds returned by base models to build a stacked ensemble #12175