def test_tweedie_distribution_power():
msg = "distribution is only defined for power<=0 and power>=1"
with pytest.raises(ValueError, match=msg):
TweedieDistribution(power=0.5)
with pytest.raises(TypeError, match="must be a real number"):
TweedieDistribution(power=1j)
with pytest.raises(TypeError, match="must be a real number"):
dist = TweedieDistribution()
dist.power = 1j
dist = TweedieDistribution()
assert isinstance(dist._lower_bound, DistributionBoundary)
assert dist._lower_bound.inclusive is False
dist.power = 1
assert dist._lower_bound.value == 0.0
assert dist._lower_bound.inclusive is True
def test_tweedie_regression_family(regression_data):
# Make sure the family attribute is always a TweedieDistribution and that
# the power attribute is properly updated
power = 2.0
est = TweedieRegressor(power=power)
assert isinstance(est.family, TweedieDistribution)
assert est.family.power == power
assert est.power == power
new_power = 0
new_family = TweedieDistribution(power=new_power)
est.family = new_family
assert isinstance(est.family, TweedieDistribution)
assert est.family.power == new_power
assert est.power == new_power
msg = "TweedieRegressor.family must be of type TweedieDistribution!"
with pytest.raises(TypeError, match=msg):
est.family = None
def test_invalid_distribution_bound():
dist = TweedieDistribution()
dist._lower_bound = 0
with pytest.raises(TypeError,
match="must be of type DistributionBoundary"):
dist.in_y_range([-1, 0, 1])
from sklearn._loss.glm_distribution import (
TweedieDistribution,
NormalDistribution,
PoissonDistribution,
GammaDistribution,
InverseGaussianDistribution,
DistributionBoundary,
)
@pytest.mark.parametrize(
"family, expected",
[
(NormalDistribution(), [True, True, True]),
(PoissonDistribution(), [False, True, True]),
(TweedieDistribution(power=1.5), [False, True, True]),
(GammaDistribution(), [False, False, True]),
(InverseGaussianDistribution(), [False, False, True]),
(TweedieDistribution(power=4.5), [False, False, True]),
],
)
def test_family_bounds(family, expected):
"""Test the valid range of distributions at -1, 0, 1."""
result = family.in_y_range([-1, 0, 1])
assert_array_equal(result, expected)
def test_invalid_distribution_bound():
dist = TweedieDistribution()
dist._lower_bound = 0
with pytest.raises(TypeError,
[
(PoissonRegressor(), True),
(GammaRegressor(), True),
(TweedieRegressor(power=1.5), True),
(TweedieRegressor(power=0), False),
],
)
def test_tags(estimator, value):
assert estimator._get_tags()["requires_positive_y"] is value
# TODO(1.3): remove
@pytest.mark.parametrize(
"est, family",
[
(PoissonRegressor(), "poisson"),
(GammaRegressor(), "gamma"),
(TweedieRegressor(), TweedieDistribution()),
(TweedieRegressor(power=2), TweedieDistribution(power=2)),
(TweedieRegressor(power=3), TweedieDistribution(power=3)),
],
)
def test_family_deprecation(est, family):
"""Test backward compatibility of the family property."""
with pytest.warns(FutureWarning, match="`family` was deprecated"):
if isinstance(family, str):
assert est.family == family
else:
assert est.family.__class__ == family.__class__
assert est.family.power == family.power
X, y, sample_weight=sample_weight_1
)
glm2 = GeneralizedLinearRegressor(**glm_params).fit(X2, y2, sample_weight=None)
assert_allclose(glm1.coef_, glm2.coef_)
@pytest.mark.parametrize("fit_intercept", [True, False])
@pytest.mark.parametrize(
"family",
[
NormalDistribution(),
PoissonDistribution(),
GammaDistribution(),
InverseGaussianDistribution(),
TweedieDistribution(power=1.5),
TweedieDistribution(power=4.5),
],
)
def test_glm_log_regression(fit_intercept, family):
"""Test GLM regression with log link on a simple dataset."""
coef = [0.2, -0.1]
X = np.array([[1, 1, 1, 1, 1], [0, 1, 2, 3, 4]]).T
y = np.exp(np.dot(X, coef))
glm = GeneralizedLinearRegressor(
alpha=0, family=family, link="log", fit_intercept=fit_intercept, tol=1e-7
)
if fit_intercept:
res = glm.fit(X[:, 1:], y)
assert_allclose(res.coef_, coef[1:], rtol=1e-6)
assert_allclose(res.intercept_, coef[0], rtol=1e-6)