hot-fixing transformation in push perturbation (#1849)

hot fixing transformation in push perturbation

Co-authored-by: Rabah Abdul Khalek <rabahabdulkhalek@Rabahs-MacBook-Pro-2.local>
Co-authored-by: Hartorn <bazire@giskard.ai>

giskard/functions/transformation.py

@@ -204,7 +204,7 @@ def mad_transformation(
     data = data.copy()
     if value_added is None:
         value_added = compute_mad(data[column_name])
-    data[column_name] = data[column_name].apply(lambda x: x + factor * value_added)
+    data[column_name] = data[column_name].apply(lambda x: x + factor * value_added).astype(data[column_name].dtype)
     return data
 
 
@@ -214,5 +214,5 @@ def add_value(data: pd.DataFrame, column_name: str, value_added: float = 0.0) ->
     Add the value_added to the column.
     """
     data = data.copy()
-    data[column_name] = data[column_name].apply(lambda x: x + value_added)
+    data[column_name] = data[column_name].apply(lambda x: x + value_added).astype(data[column_name].dtype)
     return data