Optimizing state-of-the-art feature selection methods namely SVMRFE, HSICLASSO, and mRMR with removing irrelevant and redundant features using a sparse method based on singular value decomposition (SLS) and clustering.
Three datasets, containing expression profile of healthy and Ulcerative Colitis (UC) samples, have been obtained from the Gene Expression Omnibus database(GEO). Datasets were downloaded under accession numbers GSE11223, GSE3365, and GSE22619.
- pip install pandas
- pip install mrmr (need to have Python 3.6 not higher)
- pip install pyHSICLasso