def test_max_freq_filter(self):
X = np.array([[1.0, 0.0, 0.0],
[0.0, 0.0, 0.0],
[1.0, 9.0, 0.0],
[1.0, 9.0, 0.0],
[1.0, 9.0, 0.0],
[0.0, 0.0, 0.0]])
max_freq = 0.5
filt = MaxFreqFilter(max_freq=max_freq)
Xt = filt.fit_transform(X)
# expected X
Xe = X[:,1:]
assert_array_equal(Xt, Xe)
def test_max_freq_filter_sparse(self):
X = sp.csr_matrix([[1.0, 0.0, 0.0],
[0.0, 0.0, 0.0],
[1.0, 9.0, 0.0],
[1.0, 9.0, 0.0],
[1.0, 9.0, 0.0],
[0.0, 0.0, 0.0]])
filt = MaxFreqFilter(max_freq=0.5)
Xt = filt.fit_transform(X)
# expected X
Xe = X[:,1:]
assert Xt.shape == (6,2)
# __eq__ is not properly implemented for sparse matrices
# so apply this trick
assert (Xt - Xe).nnz == 0