smazzanti / mrmr

mRMR (minimum-Redundancy-Maximum-Relevance) for automatic feature selection at scale.
MIT License
531 stars 79 forks source link

Method #31

Closed seymatakir28 closed 1 year ago

seymatakir28 commented 1 year ago

Which methods were used to assess relevance and redundancy, such as Random Forest, Mutual Information, or others?

smazzanti commented 1 year ago

They are specified in the function docstring. For instance: relevance -> "f" (f-statistic), "ks" (kolmogorov-smirnov), "rf" (random forest). redundancy -> "c" (Pearson correlation). You can also pass a callable to define your own relevance/redundancy.

seymatakir28 commented 1 year ago

So, if I don't customize it, which one does it use by default?

smazzanti commented 1 year ago

"f" for relevance and "c" for redundancy