def test_too_many_components(self):
# do not change n_components
d = table(5, 5, 1)
model = PLSRegressionLearner(n_components=4)(d)
self.assertEqual(model.skl_model.n_components, 4)
# need to use fewer components; column limited
d = table(6, 5, 1)
model = PLSRegressionLearner(n_components=6)(d)
self.assertEqual(model.skl_model.n_components, 4)
# need to use fewer components; row limited
d = table(5, 6, 1)
model = PLSRegressionLearner(n_components=6)(d)
self.assertEqual(model.skl_model.n_components, 4)
def test_too_many_components(self):
# do not change
d = table(5, 5, 1)
model = PLSRegressionLearner(n_components=5)(d)
self.assertEqual(model.skl_model.n_components, 5)
# need to use less components
d = table(6, 5, 1)
model = PLSRegressionLearner(n_components=6)(d)
self.assertEqual(model.skl_model.n_components, 5)
# number of components only depends on the number of columns
d = table(5, 6, 1)
model = PLSRegressionLearner(n_components=6)(d)
self.assertEqual(model.skl_model.n_components, 6)
def test_allow_y_dim(self):
""" The current PLS version allows only a single Y dimension. """
d = table(10, 5, 1)
learner = PLSRegressionLearner(n_components=2)
learner(d)
for n_class_vars in [0, 2]:
d = table(10, 5, n_class_vars)
with self.assertRaises(ValueError):
learner(d)
def test_compare_to_sklearn(self):
d = table(10, 5, 1)
d.X = np.random.RandomState(0).rand(*d.X.shape)
d.Y = np.random.RandomState(0).rand(*d.Y.shape)
orange_model = PLSRegressionLearner()(d)
scikit_model = PLSRegression().fit(d.X, d.Y)
np.testing.assert_almost_equal(
scikit_model.predict(d.X).ravel(), orange_model(d))
np.testing.assert_almost_equal(scikit_model.coef_,
orange_model.coefficients)
def create_learner(self):
common_args = {'preprocessors': self.preprocessors}
return PLSRegressionLearner(n_components=self.n_components,
max_iter=self.max_iter,
**common_args)