Bugfix: Only create aggregate data frame when scale is available

amazon-science · Aug 20, 2024 · c0c2590 · c0c2590
1 parent 7eebd12
commit c0c2590
Showing 1 changed file with 4 additions and 3 deletions.
diff --git a/azcausal/estimators/panel/vdid.py b/azcausal/estimators/panel/vdid.py
@@ -358,9 +358,10 @@ def vdid_panel(dx, keys, targets, time, unit, fillna=None, **kwargs):
                           .assign(delta_treat=lambda dx: dx['post_treat'] - dx['pre_treat'])
                           .assign(did=lambda dx: dx['delta_treat'] - dx['delta_contr'])
                           )
-
-        dte['cum_agg'] = dte['avg_agg'] * dte['scale']
-        dte['pct_agg'] = dte['avg_agg'].div(dte['cf'].abs() / 100, axis='index')
+        if 'scale' in dte:
+            dte['cum_agg'] = dte['avg_agg'] * dte['scale']
+        if 'cf' in dte:
+            dte['pct_agg'] = dte['avg_agg'].div(dte['cf'].abs() / 100, axis='index')
 
     return dte