ENH: Make forest prediction code more robust to mutations of the estimators list

This can be useful to quickly study the impact of the size of the forest by
subsampling from it for instance.

Author: ogrisel

sklearn/ensemble/forest.py

@@ -472,7 +472,7 @@ def predict_proba(self, X):
             for j in xrange(1, len(all_proba)):
                 proba += all_proba[j]
 
-            proba /= self.n_estimators
+            proba /= len(self.estimators_)
 
         else:
             for j in xrange(1, len(all_proba)):
@@ -572,7 +572,7 @@ def predict(self, X):
             for i in range(n_jobs))
 
         # Reduce
-        y_hat = sum(all_y_hat) / self.n_estimators
+        y_hat = sum(all_y_hat) / len(self.estimators_)
 
         return y_hat