GH-15983: Craft RuleFit Examples

[shaunyogeshwaran]

#15983
Crafted new examples for the RuleFit algorithm to showcase the different parameters.

[valenad1]

I agree with Wendy, I would also use different values than the default in the examples.

[maurever]

Why did you add examples only to the max_categorical_levels, algorithm, and max_num_rules parameters?

It would be nice to add examples also for:

• min_rule_lenght
• max_rule_length
• model_type
• distribution
• rule_generation_ntrees
• rule_importance
• predict_rules

[wendycwong]

Good call Veronika, I missed those rulefit specific parameters.

[shaunyogeshwaran]

@maurever What would some values for rule_importance and predict_rules be? The docs don't give much info

[valenad1]

@maurever What would some values for rule_importance and predict_rules be? The docs don't give much info

rule_importance is just output. So train the model and call model.rule_importance and same for predict_rules. E.g here.

[maurever]

Thanks, @valenad1, you are correct. It is my mistake to conclude them in the list of parameters. If I am correct, we can also add examples of how to use these functions. WDYT, @shaunyogeshwaran ?

[shaunyogeshwaran]

If I am correct, we can also add examples of how to use these functions. WDYT, @shaunyogeshwaran ?

Agreed, that is something I will be able to look into. @hannah-tillman what are your thoughts on this?

[hannah-tillman]

Agreed, that is something I will be able to look into. @hannah-tillman what are your thoughts on this?

@shaunyogeshwaran Yeah adding those examples in won't be an issue. Both of those calls are in written out in the gen_rulefit.py page so instead of adding them to the bottom of the examples list you'll just add them to their respective sections like this under :return:

:examples:
>>> example content
>>> example content
>>> etc
"""