def test_ridge_dtype():
x, y = datasets.make_regression(n_features=3,
n_targets=7,
n_samples=6,
random_state=1)
x = np.asarray(x, dtype="float32")
y = np.asarray(y, dtype="float32")
reg = ridge.RidgeGCV(alphas=ALPHAS).fit(x, y)
assert reg.coef_.dtype == "float32"
y = np.asarray(y, dtype="float64")
reg = ridge.RidgeGCV(alphas=ALPHAS).fit(x, y)
assert reg.coef_.dtype == "float64"
def test_simple_encoder():
x, y, voc = _dataset_and_voc()
vect = tokenization.TextVectorizer.from_vocabulary(voc)
ridge_reg = ridge.RidgeGCV().fit(x, y)
reg = ridge.FittedLinearModel(ridge_reg.coef_, ridge_reg.intercept_)
encoder = encoding.SimpleEncoder(vect, reg, mask_img=_mask_img(y.shape[1]))
text = "feature0 and feature8"
res = encoder(text)
with tempfile.TemporaryDirectory() as tmp_dir:
encoder.to_data_dir(tmp_dir)
loaded = encoding.SimpleEncoder.from_data_dir(tmp_dir)
encoded = image.get_data(loaded(text)["brain_map"])
assert np.allclose(encoded, image.get_data(res["brain_map"]))
assert len(encoder.full_vocabulary()) == 91
def test_ridge(noise, n_samples):
x, y = datasets.make_regression(
n_features=100,
n_targets=33,
effective_rank=4,
bias=100,
n_informative=7,
shuffle=False,
noise=noise,
n_samples=n_samples,
random_state=1,
)
x += 71
y -= 370
xx, yy = x.copy(), y.copy()
sk_ridge = RidgeCV(alphas=ALPHAS).fit(x, y)
reg = ridge.RidgeGCV(alphas=ALPHAS).fit(x, y)
assert sk_ridge.alpha_ == reg.alpha_
assert np.allclose(reg.coef_, sk_ridge.coef_, atol=1e-7)
assert np.allclose(reg.intercept_, sk_ridge.intercept_)
assert (x == xx).all()
assert (y == yy).all()