Add bypass attribute into column spec

FeatureExtractor excludes a column which have label or bypass
attribute.

Author: cezanne

featuretools/mkfeat/columnspec.py

@@ -116,6 +116,13 @@ def get_label_colname(self):
                 return colinfo['name']
         return None
 
+    def get_skip_colnames(self):
+        colnames = []
+        for colinfo in self.columns:
+            if ('label' in colinfo and colinfo['label']) or ('bypass' in colinfo and colinfo['bypass']):
+                colnames.append(colinfo['name'])
+        return colnames
+
     def get_is_numerics(self):
         """
         importance 결과 구성을 위하여 numeric 컬럼 여부 배열을 추출
@@ -124,7 +131,7 @@ def get_is_numerics(self):
         """
         is_numerics = []
         for colinfo in self.columns:
-            if 'label' in colinfo and colinfo['label']:
+            if 'label' in colinfo and colinfo['label'] or 'bypass' in colinfo and colinfo['bypass']:
                 is_numerics.append(False)
             else:
                 is_numerics.append(self._is_numeric_type(colinfo['type']))

featuretools/mkfeat/feat_extractor.py

@@ -90,10 +90,9 @@ def extract_features(self, operators: list) -> Error:
         self._elapsed_time.mark()
 
         self.feature_helper = FeatureHelper(features)
-        df_label = self.es.get_df_label()
-        if df_label is not None:
-            self.feature_matrix = self.feature_matrix.merge(df_label, how='left', on=df_label.columns[0])
-            self.feature_matrix.set_index(df_label.columns[0], inplace=True)
+        df_skip = self.es.get_df_skip()
+        if df_skip is not None:
+            self.feature_matrix = self.feature_matrix.join(df_skip)
 
         if self._proghandler is not None:
             self._proghandler(100)

featuretools/mkfeat/qufa_ES.py

@@ -12,7 +12,7 @@ class QufaES(EntitySet):
     def __init__(self):
         super().__init__()
         self.target_entity_name = None
-        self._df_label = None
+        self._df_skip = None
 
     def load_from_csv(self, path, colspec: ColumnSpec) -> Error:
         csv = QufaCsv(path, colspec)
@@ -21,10 +21,13 @@ def load_from_csv(self, path, colspec: ColumnSpec) -> Error:
             return data
 
         colname_key = colspec.get_key_colname()
-        colname_label = colspec.get_label_colname()
-        if colname_label:
-            self._df_label = data[[colname_key, colname_label]]
-            data = data.drop(columns=colname_label)
+        colnames_skip = colspec.get_skip_colnames()
+        if colnames_skip:
+            colnames_skip.insert(0, colname_key)
+            self._df_skip = data[colnames_skip]
+            self._df_skip.set_index(colname_key, inplace=True)
+            colnames_skip.remove(colname_key)
+            data = data.drop(columns=colnames_skip)
 
         norminfos = normalize(data, colname_key)
 
@@ -39,8 +42,8 @@ def load_from_csv(self, path, colspec: ColumnSpec) -> Error:
 
         return Error.OK
 
-    def get_df_label(self):
-        return self._df_label
+    def get_df_skip(self):
+        return self._df_skip
 
     def _search_owner_entity(self, varname):
         for et in self.entities: