BaggingClassifer/BaggingRegressor tests for sparse input

muradsalahi · muradsalahi · commit f38db565fe31 · 2014-04-17T10:20:50.000-04:00
diff --git a/sklearn/ensemble/tests/test_bagging.py b/sklearn/ensemble/tests/test_bagging.py
@@ -28,7 +28,7 @@
 from sklearn.datasets import load_boston, load_iris
 from sklearn.utils import check_random_state
 
-from scipy.sparse import csc_matrix, csr_matrix
+from scipy.sparse import csc_matrix
 
 rng = check_random_state(0)
 
@@ -81,19 +81,27 @@ def test_sparse_classification():
                           "bootstrap": [True, False],
                           "bootstrap_features": [True, False]})
 
-    for base_estimator in [DummyClassifier(),
-                           Perceptron(),
-                           KNeighborsClassifier(),
-                           SVC()]:
-        for params in grid:
-            for sparse_format in [csc_matrix, csr_matrix]:
-                X_train_sparse = sparse_format(X_train)
-                X_test_sparse = sparse_format(X_test)
-                BaggingClassifier(
-                    base_estimator=base_estimator,
-                    random_state=rng,
-                    **params
-                ).fit(X_train_sparse, y_train).predict(X_test_sparse)
+    base_estimator = SVC()
+    for params in grid:
+        sparse_format = csc_matrix
+        X_train_sparse = sparse_format(X_train)
+        X_test_sparse = sparse_format(X_test)
+
+        # Trained on sparse format
+        sparse_results = BaggingClassifier(
+            base_estimator=base_estimator,
+            random_state=check_random_state(1),
+            **params
+        ).fit(X_train_sparse, y_train).predict(X_test_sparse)
+
+        # Trained on dense format
+        dense_results = BaggingClassifier(
+            base_estimator=base_estimator,
+            random_state=check_random_state(1),
+            **params
+        ).fit(X_train, y_train).predict(X_test)
+
+        assert_array_equal(sparse_results, dense_results)
 
 
 def test_regression():
@@ -129,18 +137,27 @@ def test_sparse_regression():
                           "bootstrap": [True, False],
                           "bootstrap_features": [True, False]})
 
-    for base_estimator in [DummyRegressor(),
-                           KNeighborsRegressor(),
-                           SVR()]:
-        for params in grid:
-            for sparse_format in [csc_matrix, csr_matrix]:
-                X_train_sparse = sparse_format(X_train)
-                X_test_sparse = sparse_format(X_test)
-                BaggingRegressor(
-                    base_estimator=base_estimator,
-                    random_state=rng,
-                    **params
-                ).fit(X_train_sparse, y_train).predict(X_test_sparse)
+    base_estimator = SVR()
+    for params in grid:
+        sparse_format = csc_matrix
+        X_train_sparse = sparse_format(X_train)
+        X_test_sparse = sparse_format(X_test)
+
+        # Trained on sparse format
+        sparse_results = BaggingRegressor(
+            base_estimator=base_estimator,
+            random_state=check_random_state(1),
+            **params
+        ).fit(X_train_sparse, y_train).predict(X_test_sparse)
+
+        # Trained on dense format
+        dense_results = BaggingRegressor(
+            base_estimator=base_estimator,
+            random_state=check_random_state(1),
+            **params
+        ).fit(X_train, y_train).predict(X_test)
+
+        assert_array_equal(sparse_results, dense_results)
 
 
 def test_bootstrap_samples():
