Classification using autofeat generated features

The workflow demonstrates use of autofeat generator in classification tasks. A number of non-linear features are generated by the autofeat library. Upper panel, Random Forest, model is built using only the generated (and not the existing) features. The performance of model is comparable to the performance of model with existing features. This opens the way for building stacked models with two groups of features--existing and generated--to improve the overall predictive performance in the lower panel. Dataset used is Health Insuarnce Cross-sell data from Kaggle


This is a companion discussion topic for the original entry at https://kni.me/w/eIqCMaT1YmqX2GHo

I got an error at Autofeat Generator like the diagram at original entry.

Execute failed: The connection attempt timed out. Please consider increasing the socket timeout using the VM option ‘-Dknime.python.connecttimeout=’.