Filter Based Techniques:

Here we use statistical measures for evaluating and ranking  features independently.

  1. Provide a dataset containing  labled columns(exploratory variables) or dependent variable
  2.  Assess the feature importance using a single statistical measure (metric) like variance threshold, correlation, annova f-test, chi-square, and mutual information
  3.  Based on chosen metric system calculates a score for each feature column
  4.  Features are ranked by their scores, with the best-performing ones at the top. Decending order