Test primal-dual relationship in Ridge.

mblondel · mblondel · commit 41ae4c96acb9 · 2014-01-14T15:40:14.000+09:00
diff --git a/sklearn/linear_model/ridge.py b/sklearn/linear_model/ridge.py
@@ -108,7 +108,7 @@ def _solve_dense_cholesky(X, y, alpha):
         return coefs
 
 
-def _solve_dense_cholesky_kernel(K, y, alpha, sample_weight=None):
+def _solve_dense_cholesky_kernel(K, y, alpha, sample_weight=1.0):
     # dual_coef = inv(X X^t + alpha*Id) y
     n_samples = K.shape[0]
     n_targets = y.shape[1]
diff --git a/sklearn/linear_model/tests/test_ridge.py b/sklearn/linear_model/tests/test_ridge.py
@@ -21,6 +21,8 @@
 from sklearn.linear_model.ridge import RidgeCV
 from sklearn.linear_model.ridge import RidgeClassifier
 from sklearn.linear_model.ridge import RidgeClassifierCV
+from sklearn.linear_model.ridge import _solve_dense_cholesky
+from sklearn.linear_model.ridge import _solve_dense_cholesky_kernel
 
 
 from sklearn.cross_validation import KFold
@@ -81,6 +83,15 @@ def test_ridge():
             assert_greater(ridge.score(X, y), 0.9)
 
 
+def test_primal_dual_relationship():
+    y = y_diabetes.reshape(-1, 1)
+    coef = _solve_dense_cholesky(X_diabetes, y, alpha=[1e-2])
+    K = np.dot(X_diabetes, X_diabetes.T)
+    dual_coef = _solve_dense_cholesky_kernel(K, y, alpha=[1e-2])
+    coef2 = np.dot(X_diabetes.T, dual_coef).T
+    assert_array_almost_equal(coef, coef2)
+
+
 def test_ridge_singular():
     # test on a singular matrix
     rng = np.random.RandomState(0)
