What are the feature selection methods used to select the right variables?

 There are two basic approaches for selecting features: filter and wrapper methods.

Filtering Techniques

This entails:

• Analysis of linear discrimination
• ANOVA
• Chi-Square

The greatest analogy for feature selection is "poor data in, lousy response out." It's all about cleaning up the data coming in when we're restricting or choosing characteristics.

Wrapper Techniques

This entails:

• We test one feature at a time and keep adding them until we find a suitable match.
• Reverse Selection: We test all of the features and begin to remove them to discover which ones function best.
• Recursive Feature Elimination: Recursively examines all of the individual features and how they interact with one another.

Wrapper techniques are time-consuming, and high-end computers are required if extensive data processing is undertaken using the wrapper approach.