jwchennlp
diff --git a/‎sklearn/linear_model/coordinate_descent.py
Lines changed: 5 additions & 1 deletion b/‎sklearn/linear_model/coordinate_descent.py
Lines changed: 5 additions & 1 deletion
diff --git a/‎sklearn/linear_model/tests/test_coordinate_descent.py
Lines changed: 31 additions & 0 deletions b/‎sklearn/linear_model/tests/test_coordinate_descent.py
Lines changed: 31 additions & 0 deletions
@@ -1171,6 +1171,9 @@ class ElasticNetCV(LinearModelCV, RegressorMixin):
         all the CPUs. Note that this is used only if multiple values for
         l1_ratio are given.
 
+        positive: bool, optional
+        When set to ``True``, forces the coefficients to be positive.
+
     Attributes
     ----------
     ``alpha_`` : float
@@ -1226,7 +1229,7 @@ class ElasticNetCV(LinearModelCV, RegressorMixin):
     def __init__(self, l1_ratio=0.5, eps=1e-3, n_alphas=100, alphas=None,
                  fit_intercept=True, normalize=False, precompute='auto',
                  max_iter=1000, tol=1e-4, cv=None, copy_X=True,
-                 verbose=0, n_jobs=1):
+                 verbose=0, n_jobs=1, positive=False):
         self.l1_ratio = l1_ratio
         self.eps = eps
         self.n_alphas = n_alphas
@@ -1240,6 +1243,7 @@ def __init__(self, l1_ratio=0.5, eps=1e-3, n_alphas=100, alphas=None,
         self.copy_X = copy_X
         self.verbose = verbose
         self.n_jobs = n_jobs
+        self.positive = positive
 
 
 ###############################################################################
 
@@ -179,6 +179,22 @@ def test_lasso_cv():
     assert_greater(clf.score(X_test, y_test), 0.99)
 
 
+def test_lasso_cv_positive_constraint():
+    X, y, X_test, y_test = build_dataset()
+    max_iter = 1000
+
+    # Ensure the unconstrained fit has a negative coefficient
+    clf_unconstrained  = LassoCV(n_alphas=10, eps=1e-3, max_iter=max_iter)
+    clf_unconstrained.fit(X, y)
+    assert_true(min(clf_unconstrained.coef_) < 0)
+
+    # On same data, constrained fit has non-negative coefficients
+    clf_constrained  = LassoCV(n_alphas=10, eps=1e-3, max_iter=max_iter,
+                               positive=True)
+    clf_constrained.fit(X, y)
+    assert_true(min(clf_constrained.coef_) >= 0)
+
+
 def test_lasso_path_return_models_vs_new_return_gives_same_coefficients():
     # Test that lasso_path with lars_path style output gives the
     # same result
@@ -300,6 +316,21 @@ def test_enet_positive_constraint():
     assert_true(min(enet.coef_) >= 0)
 
 
+def test_enet_cv_positive_constraint():
+    X, y, X_test, y_test = build_dataset()
+    max_iter = 1500
+
+    # Ensure the unconstrained fit has a negative coefficient
+    enetcv_unconstrained = ElasticNetCV(max_iter=max_iter)
+    enetcv_unconstrained.fit(X, y)
+    assert_true(min(enetcv_unconstrained.coef_) < 0)
+
+    # On same data, constrained fit has non-negative coefficients
+    enetcv_constrained = ElasticNetCV(max_iter=max_iter, positive=True)
+    enetcv_constrained.fit(X, y)
+    assert_true(min(enetcv_constrained.coef_) >= 0)
+
+
 def test_multi_task_lasso_and_enet():
     X, y, X_test, y_test = build_dataset()
     Y = np.c_[y, y]