def mcfs_ours(train, test, K, debug=True):
W = mcfs(train[0], n_selected_features=K, verbose=debug)
bindices = mcfs_ranking(W)[:K]
if debug:
print(bindices)
return train[0][:, bindices], test[0][:, bindices]
def my_mcfs(X, y):
result = mcfs(copy.deepcopy(X), X.shape[1])
new_result = result.max(1)
return new_result
#!/usr/bin/env python2
# -*- coding: utf-8 -*-
import numpy as np
import scipy.io
from ConstructPairwiseDistance import ConstructPairwiseDistance
from sklearn.metrics.pairwise import pairwise_distances
import datetime
import construct_W
from skfeature.function.sparse_learning_based.MCFS import mcfs
mat = scipy.io.loadmat("COIL20.mat")
X = mat['X']
kwrags_W = {
"metric": "euclidean",
"neighbor_mode": "knn",
"weight_mode": "heat_kernel",
"k": 5,
"t": 1
}
W = construct_W.construct_W(X, **kwrags_W)
print W
weightMat = mcfs(X, 10, **{"W": W, "n_clusters": 20})
print weightMat
print weightMat.shape
np.savetxt("a.txt", weightMat, fmt='%.5f')