1

I am working on an ML problem. The dataset is of shape (15036, 216) containing all categorical variables. The task is to select the top 10 features which are applicable using any two feature selection techniques and compare those two against one another.

Surfing through internet I came across 2 algorithms which would aid me in this regard

  1. Chi-Squared Feature Selection
  2. Mutual Information Feature Selection for Classification

I did use the above two and was able to identify the top 10 relevant features. (From Chi-Square -> Based on Lower P-Values and From Mutual Information -> Based on Higher Mutual Info Values)

Now how do I compare the performance of these 2 algorithms and choose the better one amongst them?

Thiru-G
  • 11
  • 1
  • If this is an unsupervised setting, a multi-variate method would be more suitable, such as PCA. See https://stats.stackexchange.com/questions/5774/can-principal-component-analysis-be-applied-to-datasets-containing-a-mix-of-cont – msuzen Aug 20 '21 at 12:24
  • 1
    Feature selection is fraught with problems, but you’re basically looking for ways of comparing two models, regardless of how you formed them. Perhaps that will set you in the right direction. // I like [Benavoli’s JMLR paper Time for a Change](https://arxiv.org/abs/1606.04316). – Dave Aug 20 '21 at 13:36

0 Answers0