def test_rse_fit(self):
X = [[1.0, 2], [4, 8]]
X_conv = [[1.0, 2], [2, 4]]
y = [1, 2]
rse_lsq = regression.RSELinearRegression(fit_intercept=False)
lsq = linear_model.LinearRegression(fit_intercept=False)
model_rse = rse_lsq.fit(X, y)
model_lsq = lsq.fit(X_conv, np.ones(len(y)))
assert_array_equal(model_lsq.coef_, model_rse.coef_)
assert_equal(model_lsq.intercept_, model_rse.intercept_)
assert_array_almost_equal(model_rse.predict([[1, 2], [1, 2]]),
model_lsq.predict([[1, 2], [1, 2]]))
def test_multiclass(self):
X = [[1, 2], [1, 2], [4, 8], [4, 8], [200, 200], [199, 200.1],
[200.2, 198]]
y_clf = [0, 0, 0, 0, 1, 1, 1]
y_regression = [1, 1, 2, 2, 100, 100, 100]
regr_class = regression.RSELinearRegression(fit_intercept=False)
clf_class = linear_model.LogisticRegression()
multi_class = regression.MultiClassRegression(clf_class, regr_class)
model = multi_class.fit(X, y_clf, y_regression)
p = model.predict([[1, 2], [200, 200], [1, 2], [200, 200]])
assert_equal(p[0], p[2])
assert_equal(p[1], p[3])
self.assertTrue(p[0] != p[1])
def test_with_grid_search(self):
X = [[1, 2], [1, 2], [4, 8], [4, 8], [200, 200], [199, 200.1],
[200.2, 198]]
y_clf = [0, 0, 0, 0, 1, 1, 1]
y_regression = [1, 1, 2, 2, 100, 100, 100]
regr_class = GridSearchCV(regression.RSELinearRegression(),
{'normalize': [0, 1]})
clf_class = GridSearchCV(linear_model.LogisticRegression(),
{'C': [1, 2]})
multi_class = regression.MultiClassRegression(clf_class, regr_class)
model = multi_class.fit(X, y_clf, y_regression)
p = model.predict([[1, 2], [200, 200], [1, 2], [200, 200]])
assert_equal(p[0], p[2])
assert_equal(p[1], p[3])
self.assertTrue(p[0] != p[1])