Compare entities by ID in DFS

docs/source/changelog.rst

@@ -14,6 +14,7 @@ Changelog
         * Keep dataframe sorted by time during feature calculation (:pr:`626`)
         * Fix bug in encode_features that created duplicate columns of
           features with multiple outputs (:pr:`622`)
+        * Fix performance regression in DFS (:pr:`637`)
     * Changes
         * Remove unused variance_selection.py file (:pr:`613`)
         * Remove Timedelta data param (:pr:`619`)

featuretools/entityset/entity.py

@@ -117,9 +117,8 @@ def __eq__(self, other, deep=False):
             return False
         if len(self.variables) != len(other.variables):
             return False
-        for v in self.variables:
-            if v not in other.variables:
-                return False
+        if set(self.variables) != set(other.variables):
+            return False
         if deep:
             if self.last_time_index is None and other.last_time_index is not None:
                 return False

featuretools/entityset/relationship.py

@@ -108,7 +108,8 @@ def _is_unique(self):
         """Is there any other relationship with same parent and child entities?"""
         es = self.child_entity.entityset
         relationships = es.get_forward_relationships(self._child_entity_id)
-        n = len([r for r in relationships if r.parent_entity == self.parent_entity])
+        n = len([r for r in relationships
+                 if r._parent_entity_id == self._parent_entity_id])
 
         assert n > 0, 'This relationship is missing from the entityset'
 

featuretools/feature_base/feature_base.py

@@ -397,10 +397,10 @@ def __init__(self, base_feature, child_entity, relationship=None, name=None):
     def _handle_relationship(self, child_entity, relationship):
         if relationship:
             relationship_child = relationship.child_entity
-            assert child_entity == relationship_child, \
+            assert child_entity.id == relationship_child.id, \
                 'child_entity must be the relationship child entity'
 
-            assert self.parent_entity == relationship.parent_entity, \
+            assert self.parent_entity.id == relationship.parent_entity.id, \
                 'Base feature must be defined on the relationship parent entity'
         else:
             child_relationships = child_entity.entityset.get_forward_relationships(child_entity.id)
@@ -528,11 +528,11 @@ def _handle_relationship_path(self, parent_entity, relationship_path):
 
             _is_forward, first_relationship = relationship_path[0]
             first_parent = first_relationship.parent_entity
-            assert parent_entity == first_parent, \
+            assert parent_entity.id == first_parent.id, \
                 'parent_entity must match first relationship in path.'
 
             _is_forward, last_relationship = relationship_path[-1]
-            assert self.child_entity == last_relationship.child_entity, \
+            assert self.child_entity.id == last_relationship.child_entity.id, \
                 'Base feature must be defined on the entity at the end of relationship_path'
 
             path_is_unique = parent_entity.entityset \

featuretools/variable_types/variable.py

@@ -43,6 +43,9 @@ def __eq__(self, other):
             self.id == other.id and \
             self.entity_id == other.entity_id
 
+    def __hash__(self):
+        return hash((self.id, self.entity_id))
+
     def __repr__(self):
         ret = u"<Variable: {} (dtype = {})>".format(self.name, self.type_string)