Feature Selection

Feature selection, also known as variable selection or attribute selection, is a process in Machine Learning where you selectively choose those features which contribute most to your prediction variable or output in which you are interested. The main goal of this process is three-fold: improving the prediction performance of the models, providing faster and more cost-effective models, and providing a better understanding of the underlying process that generated the data.

How Feature Selection works

The feature selection process works by selecting a subset of the original features based on certain criteria like the correlation with the target, the increase in model's accuracy, or other statistical properties. Different strategies for feature selection include:

1. Filter Methods‍

These methods use statistical measures to score the relevance of individual features with respect to the target variable, like Chi-Squared Test, information gain, and correlation coefficient scores.

2. Wrapper Methods‍

These methods use a machine learning model to score the utility of a subset of features, which means the search for the best feature subset is "wrapped" around the algorithm, like recursive feature elimination.

3. Embedded Methods‍

These methods perform feature selection as a part of the model construction process, like LASSO and Ridge regression.

Feature selection can give you a boost in accuracy by both enabling the algorithms to train faster (as there are fewer data to process) and reducing overfitting (by removing redundant features).