# of deletions rel to initial message length

change_visualizer.py

@@ -82,7 +82,7 @@ if __name__ == '__main__':
             def layout_for_statistics(statistics_group_name):
                 gr.Markdown(f"### {statistics_group_name}")
                 stats = STATISTICS[statistics_group_name]
-                gr.Number(label="Average deletions number (rel to the result length)", interactive=False,
+                gr.Number(label="Average deletions number (rel to the initial message length)", interactive=False,
                           value=stats['deletions'].mean().item(), precision=3)
                 gr.Number(label="Average insertions number (rel to the result length)", interactive=False,
                           value=stats['insertions'].mean().item(), precision=3)

statistics.py

@@ -2,7 +2,7 @@ import numpy as np
 import pandas as pd
 
 
-def get_statistics_for_df(df: pd.DataFrame, end_col, annotated_col):
+def get_statistics_for_df(df: pd.DataFrame, start_col, end_col, annotated_col):
     relative_deletions = []
     relative_insertions = []
     relative_changes = []
@@ -18,8 +18,9 @@ def get_statistics_for_df(df: pd.DataFrame, end_col, annotated_col):
 
         sum_changes = sum_deletions + sum_insertions
         end_length = len(row[end_col])
+        start_length = len(row[start_col])
 
-        relative_deletions.append(sum_deletions / end_length)
+        relative_deletions.append(sum_deletions / start_length)
         relative_insertions.append(sum_insertions / end_length)
         relative_changes.append(sum_changes / end_length)
 
@@ -31,8 +32,9 @@ def get_statistics_for_df(df: pd.DataFrame, end_col, annotated_col):
 
 
 def get_statistics_for_manual_df(df):
-    return get_statistics_for_df(df, end_col='commit_msg_end', annotated_col='annotated_diff')
+    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, end_col='reference', annotated_col='annotated_diff')
+    return get_statistics_for_df(df, start_col="commit_msg_start", end_col='reference', annotated_col='annotated_diff')