By @rpinho

Refereed correctness bugs in open-source research software, with patches and links to upstream PRs

c4bb77c / feedzai-deccaf/patches/DECCAF-001.patch

Raw
diff --git a/Code/deferral/run_alert.py b/Code/deferral/run_alert.py
index ae608b6..b18f73c 100644
--- a/Code/deferral/run_alert.py
+++ b/Code/deferral/run_alert.py
@@ -72,7 +72,7 @@ def rand_deferral_func(capacities, batches, testset, expert_preds, env, model, s
     results = pd.DataFrame(columns = ['prediction'])
     for i in np.arange(1,batches['batch'].max()+1):
         c = capacities.loc[capacities['batch_id'] == i,:].iloc[0]
-        c.loc['model#0'] = c['batch_size'] - c[2:].sum()
+        c.loc['model#0'] = c['batch_size'] - c.drop(labels=['batch_id', 'batch_size', 'model#0'], errors='ignore').sum()
         cases = testset.loc[batches.loc[batches['batch'] == i]['case_id'],:]
         experts = expert_preds.columns.to_list()
         for ix in cases.index:
@@ -107,7 +107,7 @@ def ova_deferral_func(capacities, batches, testset, expert_preds, model_preds, e
     results = pd.DataFrame(columns = ['prediction'])
     for i in np.arange(1,batches['batch'].max()+1):
         c = capacities.loc[capacities['batch_id'] == i,:].iloc[0]
-        c.loc['model#0'] = c['batch_size'] - c[2:].sum()
+        c.loc['model#0'] = c['batch_size'] - c.drop(labels=['batch_id', 'batch_size', 'model#0'], errors='ignore').sum()
         cases = testset.loc[batches.loc[batches['batch'] == i]['case_id'],:]
         preds = model_preds.loc[cases.index]
         for ix, row in preds.iterrows():
@@ -139,7 +139,7 @@ def deccaf_cp_deferral_func(capacities, batches, testset, expert_preds, model_pr
     results = pd.DataFrame(columns = ['prediction'])
     for b in np.arange(1,batches['batch'].max()+1):
         c = capacities.loc[capacities['batch_id'] == b,:].iloc[0]
-        c.loc['model#0'] = c['batch_size'] - c[2:].sum()
+        c.loc['model#0'] = c['batch_size'] - c.drop(labels=['batch_id', 'batch_size', 'model#0'], errors='ignore').sum()
         cases = testset.loc[batches.loc[batches['batch'] == b]['case_id'],:] 
         preds = model_preds.loc[cases.index]
         cost_matrix_df = preds.T