Reverting using the same DF in process_columns

dagardner-nv · Sep 13, 2022 · 41fcf09 · 41fcf09
1 parent 23f493b
commit 41fcf09
Showing 1 changed file with 3 additions and 2 deletions.
diff --git a/examples/dfp_workflow/morpheus/dfp/utils/column_info.py b/examples/dfp_workflow/morpheus/dfp/utils/column_info.py
@@ -112,13 +112,14 @@ def __post_init__(self):
 
 def _process_columns(df_in: pd.DataFrame, input_schema: DataFrameInputSchema):
 
-    output_df = df_in
+    # TODO(MDD): See what causes this to have such a perf impact over using df_in
+    output_df = pd.DataFrame()
 
     # Iterate over the column info
     for ci in input_schema.column_info:
         output_df[ci.name] = ci.process_column(df_in)
 
-    if (False and input_schema.preserve_columns is not None):
+    if (input_schema.preserve_columns is not None):
         # Get the list of remaining columns not already added
         df_in_columns = set(df_in.columns) - set(output_df.columns)