ENH add support for different feature names in HFL (primihub#777)

* ENH add support for different feature names in HFL * use get for dictionary
phoenix20162016 · May 28, 2024 · 97b40f5 · 97b40f5
1 parent daea4dd
commit 97b40f5
Show file tree

Hide file tree

Showing 4 changed files with 14 additions and 15 deletions.
diff --git a/python/primihub/FL/linear_regression/hfl_client.py b/python/primihub/FL/linear_regression/hfl_client.py
@@ -47,7 +47,9 @@ def train(self):
                                     task_info=self.task_info)
 
         # load dataset
-        selected_column = self.common_params['selected_column']
+        selected_column = self.common_params.get('selected_column')
+        if selected_column is None:
+            selected_column = self.role_params.get('selected_column')
         id = self.common_params['id']
         x = read_data(data_info=self.role_params['data'],
                       selected_column=selected_column,

diff --git a/python/primihub/FL/logistic_regression/hfl_client.py b/python/primihub/FL/logistic_regression/hfl_client.py
@@ -48,7 +48,9 @@ def train(self):
                                     task_info=self.task_info)
 
         # load dataset
-        selected_column = self.common_params['selected_column']
+        selected_column = self.common_params.get('selected_column')
+        if selected_column is None:
+            selected_column = self.role_params.get('selected_column')
         id = self.common_params['id']
         x = read_data(data_info=self.role_params['data'],
                       selected_column=selected_column,

diff --git a/python/primihub/FL/neural_network/hfl_client.py b/python/primihub/FL/neural_network/hfl_client.py
@@ -46,7 +46,9 @@ def train(self):
                                     task_info=self.task_info)
 
         # load dataset
-        selected_column = self.common_params['selected_column']
+        selected_column = self.common_params.get('selected_column')
+        if selected_column is None:
+            selected_column = self.role_params.get('selected_column')
         id = self.common_params['id']
         x = read_data(data_info=self.role_params['data'],
                       selected_column=selected_column,

diff --git a/python/primihub/FL/preprocessing/pipeline.py b/python/primihub/FL/preprocessing/pipeline.py
@@ -49,7 +49,9 @@ def fit_transform(self):
 
         # load dataset
         if FL_type == "H":
-            selected_column = self.common_params["selected_column"]
+            selected_column = self.common_params.get('selected_column')
+            if selected_column is None:
+                selected_column = self.role_params.get('selected_column')
             id = self.common_params["id"]
         else:
             selected_column = self.role_params["selected_column"]
@@ -91,6 +93,8 @@ def fit_transform(self):
         # preprocessing
         if FL_type == "H":
             module_params = self.common_params.get("preprocess_module")
+            if module_params is None:
+                module_params = self.role_params.get("preprocess_module")
         else:
             module_params = self.role_params.get("preprocess_module")
             if module_params is None:
@@ -130,21 +134,10 @@ def fit_transform(self):
                     elif "Scaler" in module_name:
                         column = data[column].select_dtypes(include=num_type).columns
 
-                if role == "client":
-                    channel.send("column", column)
-                    column = channel.recv("column")
-                if role == "server":
-                    client_column = channel.recv_all("column")
-                    column = list(set(chain.from_iterable(client_column)))
-                    channel.send_all("column", column)
-
             if isinstance(column, pd.Index):
                 column = column.tolist()
             if column:
                 logger.info(f"column: {column}, # {len(column)}")
-            else:
-                logger.info(f"column is empty, {module_name} is skipped")
-                continue
 
             module = select_module(module_name, params, FL_type, role, channel)