mrmr
mrmr copied to clipboard
smazzanti
mRMR (minimum-Redundancy-Maximum-Relevance) for automatic feature selection at scale.
fucntions retruend all NaN score of redundancy even though relevance score is not NaN.
https://github.com/smazzanti/mrmr/blob/main/LICENSE is the GPL-3.0 license, but https://github.com/smazzanti/mrmr/blob/main/setup.py identifies the license as MIT. Can you clarify what license actually applies to this package? Thanks in advance.
Dear authors, I want to know how can I find the optimum number of features in gridsearchCV ?
Version 0.2.4 introduces this bug: ```TypeError: random_forest_classif() got an unexpected keyword argument 'n_jobs'``` Could you please have a look?
I used a pre-trained vgg16 model and extracted the features from some of the layers. Later i passed the features from GAP layer and concatenate it now i want to...
I'm thinking of using mrmr with a nested cross validation pipeline, do you have any suggestions on how to combine/aggregate the resulting selected features and ranking for each fold?
@smazzanti, thank you for this package and the Medium article explaining MRMR in a very good and clear way. Hope this package can only increase and improve, and hopefully I...
hi @smazzanti , I just find out this package in your Medium article, thank you for develope the mrmr package for slove the regression problem and the aritcle in Medium...
Is it possible to additionally return the final score used for sorting? I would like to make the number of features actually used dependent on a threshold on this score....
Hi, it seems that selected features are not ordered by relevance : selected_features, relevance, redundacy = outputs_from_mrmr print(relevance.values[selected_features[0:5]]) [77.04191317 7.65999306 2.1369512 71.32355801 44.90570244] Is it a bug or the user...
