Update app.py

app.py

@@ -1272,10 +1272,17 @@ with tab_predict:
     pipe = st.session_state.pipe
     meta = st.session_state.meta
 
-    st.write("DEBUG: feature columns as loaded from meta:")
-    st.write(feature_cols)
+    # 1) MUST come first: schema from meta
+    feature_cols = meta["schema"]["features"]
+    num_cols = meta["schema"]["numeric"]
+    cat_cols = meta["schema"]["categorical"]
 
+    # 2) Now we can build lookup
+    FEATURE_LOOKUP = {norm_col(c): c for c in feature_cols}
 
+    # 3) Debug (NOW valid)
+    st.write("DEBUG: feature columns as loaded from meta:")
+    st.write(feature_cols)
 
     from datetime import date
 
@@ -1544,6 +1551,9 @@ with tab_predict:
     num_cols = meta["schema"]["numeric"]
     cat_cols = meta["schema"]["categorical"]
 
+    st.write("DEBUG: feature columns as loaded from meta:")
+    st.write(feature_cols)
+
     # Map normalized name -> actual model column name
     FEATURE_LOOKUP = {norm_col(c): c for c in feature_cols}
 
@@ -1820,7 +1830,9 @@ with tab_predict:
     
             out = X_one.copy()
             out["predicted_probability"] = proba_one
+            pred_class = int(proba_one >= thr_single)
             out["predicted_class"] = pred_class
+            
             out["risk_band"] = band_one(proba_one)
         
             st.dataframe(out, use_container_width=True)