[MRG+2] FIX IsolationForest(max_features=0.8).predict(X) fails input validation (scikit-learn#5757)

Fixes scikit-learn#5732

Author: IshankGulati

doc/whats_new.rst

@@ -136,6 +136,10 @@ Bug fixes
    - Fix estimators to accept a ``sample_weight`` parameter of type
      ``pandas.Series`` in their ``fit`` function. :issue:`7825` by
      `Kathleen Chen`_.
+  
+   - Fixed a bug where :class:`sklearn.ensemble.IsolationForest` fails when 
+     ``max_features`` is less than 1.
+     :issue:`5732` by :user:`Ishank Gulati <IshankGulati>`.
 
    - Fix a bug where :class:`sklearn.ensemble.VotingClassifier` raises an error
      when a numpy array is passed in for weights. :issue:`7983` by

sklearn/ensemble/iforest.py

@@ -248,17 +248,28 @@ def decision_function(self, X):
         """
         # code structure from ForestClassifier/predict_proba
         # Check data
-        X = self.estimators_[0]._validate_X_predict(X, check_input=True)
+        X = check_array(X, accept_sparse='csr')
         n_samples = X.shape[0]
 
         n_samples_leaf = np.zeros((n_samples, self.n_estimators), order="f")
         depths = np.zeros((n_samples, self.n_estimators), order="f")
 
-        for i, tree in enumerate(self.estimators_):
-            leaves_index = tree.apply(X)
-            node_indicator = tree.decision_path(X)
+        if self._max_features == X.shape[1]:
+            subsample_features = False
+        else:
+            subsample_features = True
+
+        for i, (tree, features) in enumerate(zip(self.estimators_,
+                                                 self.estimators_features_)):
+            if subsample_features:
+                X_subset = X[:, features]
+            else:
+                X_subset = X
+            leaves_index = tree.apply(X_subset)
+            node_indicator = tree.decision_path(X_subset)
             n_samples_leaf[:, i] = tree.tree_.n_node_samples[leaves_index]
-            depths[:, i] = np.asarray(node_indicator.sum(axis=1)).reshape(-1) - 1
+            depths[:, i] = np.ravel(node_indicator.sum(axis=1))
+            depths[:, i] -= 1
 
         depths += _average_path_length(n_samples_leaf)
 

sklearn/ensemble/tests/test_iforest.py

@@ -200,3 +200,14 @@ def test_max_samples_consistency():
     X = iris.data
     clf = IsolationForest().fit(X)
     assert_equal(clf.max_samples_, clf._max_samples)
+
+
+def test_iforest_subsampled_features():
+    # It tests non-regression for #5732 which failed at predict.
+    rng = check_random_state(0)
+    X_train, X_test, y_train, y_test = train_test_split(boston.data[:50],
+                                                        boston.target[:50],
+                                                        random_state=rng)
+    clf = IsolationForest(max_features=0.8)
+    clf.fit(X_train, y_train)
+    clf.predict(X_test)