feature-selection

Three feature selection methods including variance threshold, correlation threshold and forward search. Tested on data from sci-kit learn library.

1. Variance threshold: Removes features whose variance falls below a threshold.
2. Correlation threshold: Removes features whose correlation falls below a threshold. Uses Pearson's correlation.
3. Forward search Wrapper-based method, basically brute-force. In each step it finds the best performing feature for the current set.

Datasets: Iris (4 features) Wine (13 features) Breast cancer (30 features)