def _test_input_format_f_contiguous_numpy(dtype):
rng = np.random.RandomState(0)
x_default = np.array(5 * rng.random_sample((10, 4)), dtype=dtype)
x_numpy = np.asanyarray(x_default, dtype=dtype, order='F')
assert not x_numpy.flags.c_contiguous
assert x_numpy.flags.f_contiguous
assert x_numpy.flags.fnc
expected = linear_kernel(x_default)
result = linear_kernel(x_numpy)
assert_allclose(expected, result)
def _test_input_format_c_contiguous_pandas(dtype):
pd = pytest.importorskip('pandas')
rng = np.random.RandomState(0)
x_default = np.array(5 * rng.random_sample((10, 4)), dtype=dtype)
x_numpy = np.asanyarray(x_default, dtype=dtype, order='C')
assert x_numpy.flags.c_contiguous
assert not x_numpy.flags.f_contiguous
assert not x_numpy.flags.fnc
x_df = pd.DataFrame(x_numpy)
expected = linear_kernel(x_df)
result = linear_kernel(x_numpy)
assert_allclose(expected, result)
def test_dense_self_linear_kernel():
rng = np.random.RandomState(0)
X = np.array(5 * rng.random_sample((10, 4)))
result = linear_kernel(X)
expected = np.dot(X, np.array(X).T)
assert_allclose(result, expected, rtol=1e-15)
def _test_input_format_c_not_contiguous_numpy(dtype):
rng = np.random.RandomState(0)
x_default = np.array(5 * rng.random_sample((10, 4)), dtype=dtype)
dummy_data = np.insert(x_default, range(1, x_default.shape[1]), 8, axis=1)
x_numpy = dummy_data[:, ::2]
assert_allclose(x_numpy, x_default)
assert not x_numpy.flags.c_contiguous
assert not x_numpy.flags.f_contiguous
assert not x_numpy.flags.fnc
expected = linear_kernel(x_default)
result = linear_kernel(x_numpy)
assert_allclose(expected, result)
def _test_dense_small_linear_kernel(scale, shift, dtype):
rng = np.random.RandomState(0)
X = np.array(5 * rng.random_sample((10, 4)), dtype=dtype)
Y = np.array(5 * rng.random_sample((15, 4)), dtype=dtype)
result = linear_kernel(X, Y, scale=scale, shift=shift)
expected = np.dot(X, np.array(Y).T) * scale + shift
tol = 1e-14 if dtype == np.float64 else 1e-6
assert_allclose(result, expected, rtol=tol)