def test_LinearRegression_warm_start(self):
"""...Test LinearRegression warm start
"""
X, y, weights0, intercept0 = Test.get_train_data()
fit_intercepts = [True, False]
cases = itertools.product(LinearRegression._solvers.keys(),
fit_intercepts)
for solver, fit_intercept in cases:
solver_kwargs = {
'solver': solver,
'max_iter': 2,
'fit_intercept': fit_intercept,
'warm_start': True,
'tol': 0
}
learner = LinearRegression(**solver_kwargs)
learner.fit(X, y)
if fit_intercept:
coeffs_1 = np.hstack((learner.weights, learner.intercept))
else:
coeffs_1 = learner.weights
learner.fit(X, y)
if fit_intercept:
coeffs_2 = np.hstack((learner.weights, learner.intercept))
else:
coeffs_2 = learner.weights
# Thanks to warm start objective should have decreased
self.assertLess(
learner._solver_obj.objective(coeffs_2),
learner._solver_obj.objective(coeffs_1))
def test_LinearRegression_fit(self):
"""...Test LinearRegression fit with different solvers and penalties
"""
fit_intercepts = [False, True]
n_samples = 2000
n_features = 20
X, y, weights0, _ = self.get_train_data(
n_samples=n_samples, n_features=n_features, fit_intercept=False)
intercept0 = -1
for i, (solver, penalty, fit_intercept) \
in enumerate(product(LinearRegression._solvers.keys(),
LinearRegression._penalties.keys(),
fit_intercepts)):
if fit_intercept:
y_ = y + intercept0
else:
y_ = y.copy()
if penalty == 'binarsity':
learner = LinearRegression(
verbose=False, fit_intercept=fit_intercept, solver=solver,
penalty=penalty, tol=1e-10, blocks_start=[0],
blocks_length=[1])
else:
learner = LinearRegression(
verbose=False, fit_intercept=fit_intercept, solver=solver,
penalty=penalty, tol=1e-10)
learner.fit(X, y_)
err = norm(learner.weights - weights0) / n_features ** 0.5
self.assertLess(err, 3e-2)
if fit_intercept:
self.assertLess(abs(learner.intercept - intercept0), 3e-2)
def test_score(self):
"""...Test LinearRegression predict
"""
X_train, y_train, _, _ = self.get_train_data(n_samples=2000,
n_features=12)
learner = LinearRegression(random_state=32789, tol=1e-9)
learner.fit(X_train, y_train)
X_test, y_test, _, _ = self.get_train_data(n_samples=200,
n_features=12)
self.assertAlmostEqual(
learner.score(X_test, y_test), 0.793774, places=4)
def test_predict(self):
"""...Test LinearRegression predict
"""
X_train, y_train, _, _ = self.get_train_data(n_samples=200,
n_features=12)
learner = LinearRegression(random_state=32789, tol=1e-9)
learner.fit(X_train, y_train)
X_test, y_test, _, _ = self.get_train_data(n_samples=5, n_features=12)
y_pred = np.array([0.084, -1.4276, -3.1555, 2.6218, 0.3736])
np.testing.assert_array_almost_equal(
learner.predict(X_test), y_pred, decimal=4)
