SPEC¶

skfeature.function.similarity_based.SPEC

Description¶

SPEC (Spectral Feature Selection) is a spectral method that can be used in both supervised and unsupervised settings. It scores features by their consistency with the eigenvectors of a similarity (affinity) matrix of the samples.

Usage¶

import numpy as np
from sklearn.datasets import load_iris

from skfeature.function.similarity_based import SPEC

X, y = load_iris(return_X_y=True)

# get a score for every feature (aligned with SelectKBest)
score = SPEC.spec(X, y)

# or get the indices of the selected features
selected = SPEC.spec(X, y, mode="index")

Parameters¶

X: numpy array, shape (n_samples, n_features) — input data
y: numpy array, shape (n_samples,) or None — optional class labels
**kwargs: additional parameters (e.g. n_neighbors for the similarity graph)
mode: {{"rank", "index"}}, default "rank" — "rank" returns an array of feature indices ordered by importance and aligned with sklearn.feature_selection.SelectKBest; "index" returns the indices of the selected features with the most important one first

Returns¶

score: numpy array, shape (n_features,) — ranking score of every feature, aligned with sklearn.feature_selection.SelectKBest

References¶

Original implementation from the DMML Lab@ASU Feature Selection Repository.