Fix the statistics in visualization

change_visualizer.py

@@ -9,8 +9,8 @@ n_diffs_manual = len(df_manual)
 df_synthetic = generate_annotated_diffs.synthetic_data_with_annotated_diffs()
 n_diffs_synthetic = len(df_synthetic)
 
-STATISTICS = {"manual": statistics.get_statistics_for_manual_df(df_manual),
-              "synthetic": statistics.get_statistics_for_synthetic_df(df_synthetic)}
+STATISTICS = {"manual": statistics.get_statistics_for_df(df_manual),
+              "synthetic": statistics.get_statistics_for_df(df_synthetic)}
 
 
 def update_dataset_view(diff_idx):

statistics.py

@@ -22,18 +22,10 @@ def get_statistics(start_msg, end_msg, annotated_msg):
     }
 
 
-def get_statistics_for_df(df: pd.DataFrame, start_col, end_col, annotated_col):
-    stats = [get_statistics(row[start_col], row[end_col], row[annotated_col]) for _, row in df.iterrows()]
+def get_statistics_for_df(df: pd.DataFrame):
+    stats = [get_statistics(row["commit_msg_start"], row["commit_msg_end"], row["annotated_diff"]) for _, row in
+             df.iterrows()]
 
     assert len(stats) > 0
 
     return {stat_name: np.asarray([e[stat_name] for e in stats]) for stat_name in stats[0]}
-
-
-def get_statistics_for_manual_df(df):
-    return get_statistics_for_df(df, start_col="commit_msg_start", end_col='commit_msg_end',
-                                 annotated_col='annotated_diff')
-
-
-def get_statistics_for_synthetic_df(df):
-    return get_statistics_for_df(df, start_col="initial_msg_pred", end_col='reference', annotated_col='annotated_diff')