def test_tfidf_transformer_sparse():
X = cupyx.scipy.sparse.rand(10, 2000, dtype=np.float64, random_state=123)
X_csc = cupyx.scipy.sparse.csc_matrix(X)
X_csr = cupyx.scipy.sparse.csr_matrix(X)
X_trans_csc = TfidfTransformer().fit_transform(X_csc).todense()
X_trans_csr = TfidfTransformer().fit_transform(X_csr).todense()
cp.testing.assert_array_almost_equal(X_trans_csc, X_trans_csr)
def test_tfidf_transformer_copy(norm, use_idf, smooth_idf, sublinear_tf):
if use_idf:
pytest.xfail(
"cupyx.scipy.sparse.csr does not support inplace multiply.")
data_gpu = cupyx.scipy.sparse.csr_matrix(
cp.array([[0, 1, 1, 1], [0, 2, 0, 1]], dtype=cp.float64, order='F'))
tfidf = TfidfTransformer(norm=norm,
use_idf=use_idf,
smooth_idf=smooth_idf,
sublinear_tf=sublinear_tf)
res = tfidf.fit_transform(data_gpu, copy=False)
cp.testing.assert_array_almost_equal(data_gpu.todense(), res.todense())
def test_tfidf_transformer(data, norm, use_idf, smooth_idf, sublinear_tf):
data_gpu = cp.array(data)
tfidf = TfidfTransformer(norm=norm,
use_idf=use_idf,
smooth_idf=smooth_idf,
sublinear_tf=sublinear_tf)
sk_tfidf = SkTfidfTransfo(norm=norm,
use_idf=use_idf,
smooth_idf=smooth_idf,
sublinear_tf=sublinear_tf)
res = tfidf.fit_transform(data_gpu).todense()
ref = sk_tfidf.fit_transform(data).todense()
cp.testing.assert_array_almost_equal(res, ref)