-
Add Rulefit MOJO in a similar pattern as it’s done for KLime or Stacked Ensemble (RF MOJO has a reference on GLM MOJO …)
-
-
Rulefit returns rules with splits that are unnecessary. For example, x < 3 & x < 5.
-
to have better control over regularization strength
-
The method `_extract_rules()` in the class `RuleFit` has `List[Rule]` as type annotation for the return value.
However it returns the result of `extract_rulefit()` which returns a list of strings.
-
References:
* [http://statweb.stanford.edu/~jhf/ftp/RuleFit.pdf|http://statweb.stanford.edu/~jhf/ftp/RuleFit.pdf]
* [https://statweb.stanford.edu/~jhf/R_RuleFit.html|https://statweb.stanford.edu/~jhf…
-
Issue + logs described here: [https://support.h2o.ai/a/tickets/101619|https://support.h2o.ai/a/tickets/101619|smart-link]
{noformat}java.lang.OutOfMemoryError: GC overhead limit exceeded
a…
-
-
Python example on [http://docs.h2o.ai/h2o/latest-stable/h2o-docs/data-science/rulefit.html|http://docs.h2o.ai/h2o/latest-stable/h2o-docs/data-science/rulefit.html] has
{{from h2o.estimators import H2O…
-
When running rulefit, there can be multiple rules that are exactly the same. We should remove these as a post-processing step.