H2O Generalized Linear Model for regression

KCH Workflows
1

This example shows how to build an H2O GLM model for regression, predict new data and score the regression metrics for model evaluation.

This is a companion discussion topic for the original entry at https://kni.me/w/FdAAnJjLcR0NjMPe
2

Hi, I can’t work out how to see the P-Values in the coefficients table - any ideas?

3

Also, the H2O documentation suggests you can fit interactions between variables but I can’t see any option in the configuration - is there a way of doing this that doesn’t involve Python or R?

4

Hi @marisamurton,

Since p-values are only provided by h2o for very restricted parameter settings (see documentation), we don’t output the p values a the moment.

This is currently not supported by the node. Just fyi, h2o computes the interactions on the fly if the parameter is not set. See interactions — H2O 3.44.0.3 documentation.

Best,
Simon

5

Thanks for replying Simon - I’m glad it wasn’t me being thick. The package I was used to for GLMs (Emblem) shows p values and allows easy fitting of interactions, they were useful features but alas the license fee is expensive…