Util

skfeature.utility.util

Description

Util contains general-purpose helpers shared across the repository, most importantly the rank conversion used to align feature rankings with scikit-learn's SelectKBest interface.

Usage

import numpy as np

from skfeature.utility import util

indices = np.array([3, 1, 2, 0])  # 0 is the most important feature
rank = util.reverse_argsort(indices, size=4)

Functions

• reverse_argsort(X, size=None): convert feature indices (0 = most important) into a rank array aligned with sklearn.feature_selection.SelectKBest
• loadmat(path_to_mat): load a MATLAB .mat file into a dictionary of numpy arrays