Memory scalability issues in univariate feature selection · scikit-learn/scikit-learn#1651

(8 comments) (0 reactions) (0 assignees)Python (66,084 stars) (27,020 forks)batch import

EnhancementPerformancehelp wantedmodule:feature_selection

説明

Gensim has a re-implementation of selectkbest to fix memory scalability issues. Someone should review them and it integrate the fix back into the original sklearn codebase.

https://github.com/piskvorky/gensim/blob/a73c84e21aecd3cc77ba2d752912f73b712bc60a/gensim/models/selectkbest.py

Maybe @agramfort you can have a look?

コントリビューターガイド

技術スタック: python
領域: machine learning
Issue 種別: feature
難度: 4
推定時間: over 1 week
活動状況: stale
明確さ: needs investigation
前提条件: Pythonnumpyscikit learn internals
初心者向け度: 20
調査方針: Review the gensim re implementation of selectkbest at https://github.com/piskvorky/gensim/blob/a73c84e21aecd3cc77ba2d752912f73b712bc60a/gensim/models/selectkbest.py and understand its memory efficient approach. Then identify the relevant code in scikit learn's feature selection module (likely in sklearn/feature selection) and propose an integration that fixes memory scalability without breaking existing functionality. Check issue comments for any prior discussion or maintainer feedback.