BUG: catch any/all case that fails in NumPy > 1.6 with stricter casting rules

Author: wesm

pandas/core/frame.py

@@ -4585,9 +4585,7 @@ def _reduce(self, op, axis=0, skipna=True, numeric_only=None,
                     result = result.astype(np.float64)
                 elif filter_type == 'bool' and notnull(result).all():
                     result = result.astype(np.bool_)
-                else:
-                    raise NotImplementedError
-
+                # otherwise, accept it
             except (ValueError, TypeError):
                 pass