Feature selection is a crucial problem to tackle. Having too many features may lead to high training time and sub-optimal performence. This repository means to supplement the comprehensive blog post I wrote about the different methods for feature selection and was published here.
In the blog post, I present various methods that are based on different approaches, like correlations, information theory and distance.
Different feature selectors are available for Python via different packages, some of which need to be better maintained. Therefore, in the current repo, I put them together in one place, fixing bugs and preparing a convenient wrapper two wrap them all in one class for convenient inference.