Update app.py

app.py

@@ -96,10 +96,19 @@ def train_and_save(df: pd.DataFrame):
     X = df[FEATURE_COLS].copy()
     y_raw = df[LABEL_COL].copy()
 
-    for c in NUM_COLS:
-        X[c] = pd.to_numeric(X[c], errors="coerce")
-    for c in CAT_COLS:
-        X[c] = X[c].astype("string")
+# Replace pandas NA with numpy nan globally (critical for sklearn)
+X = X.replace({pd.NA: np.nan})
+
+# Force numeric columns to float with NaNs
+for c in NUM_COLS:
+    X[c] = pd.to_numeric(X[c], errors="coerce")
+
+# Force categorical columns to plain object strings, but keep missing as np.nan (not pd.NA)
+for c in CAT_COLS:
+    X[c] = X[c].astype("object")
+    X.loc[X[c].isna(), c] = np.nan
+
+
 
     y01, pos_class = coerce_binary_label(y_raw)
 
@@ -216,11 +225,13 @@ with tab_predict:
         if infer_file:
             df_inf = pd.read_excel(infer_file, engine="openpyxl")
             X_inf = df_inf[FEATURE_COLS].copy()
+            X_inf = X_inf.replace({pd.NA: np.nan})
+
 
             for c in NUM_COLS:
                 X_inf[c] = pd.to_numeric(X_inf[c], errors="coerce")
             for c in CAT_COLS:
-                X_inf[c] = X_inf[c].astype("string")
+                X_inf[c] = X_inf[c].astype("object")
 
             pipe = st.session_state.pipe
             proba = pipe.predict_proba(X_inf)[:, 1]