Update inference.py

inference.py

@@ -9,10 +9,13 @@ import warnings
 
 warnings.filterwarnings("ignore", category=UserWarning, module="sklearn")
 
+print("===== Application Startup =====")
+print(f"Working directory: {os.getcwd()}")
+print(f"Files present: {os.listdir('.')}")
+
 # ---------------------------------------------------------------------------
 # Compatibility patch — inject _RemainderColsList if the installed sklearn
-# version does not have it (added in sklearn 1.4+). This allows .skops files
-# saved with a newer sklearn to load correctly on older environments.
+# version does not have it (added in sklearn 1.4+).
 # ---------------------------------------------------------------------------
 import sklearn.compose._column_transformer as _ct
 if not hasattr(_ct, "_RemainderColsList"):
@@ -24,6 +27,7 @@ if not hasattr(_ct, "_RemainderColsList"):
     _ct._RemainderColsList = _RemainderColsList
     import sklearn.compose
     sklearn.compose._RemainderColsList = _RemainderColsList
+    print("Patched _RemainderColsList into sklearn.compose")
 
 
 # ---------------------------------------------------------------------------
@@ -32,21 +36,9 @@ if not hasattr(_ct, "_RemainderColsList"):
 
 NUM_COLUMNS = ["AGE", "NACS2YR"]
 CATEG_COLUMNS = [
-    "AGEGPFF",
-    "SEX",
-    "KPS",
-    "DONORF",
-    "GRAFTYPE",
-    "CONDGRPF",
-    "CONDGRP_FINAL",
-    "ATGF",
-    "GVHD_FINAL",
-    "HLA_FINAL",
-    "RCMVPR",
-    "EXCHTFPR",
-    "VOC2YPR",
-    "VOCFRQPR",
-    "SCATXRSN",
+    "AGEGPFF", "SEX", "KPS", "DONORF", "GRAFTYPE", "CONDGRPF",
+    "CONDGRP_FINAL", "ATGF", "GVHD_FINAL", "HLA_FINAL",
+    "RCMVPR", "EXCHTFPR", "VOC2YPR", "VOCFRQPR", "SCATXRSN",
 ]
 
 FEATURE_NAMES = NUM_COLUMNS + CATEG_COLUMNS
@@ -82,23 +74,29 @@ DEFAULT_N_BOOT_CI   = 500
 # ---------------------------------------------------------------------------
 
 def _load_skops_model(fname):
+    """Load a skops model file. Raises RuntimeError on failure (no sys.exit)."""
+    if not os.path.exists(fname):
+        raise RuntimeError(f"Model file not found: {fname}")
     try:
         untrusted = sio.get_untrusted_types(file=fname)
-        return sio.load(fname, trusted=untrusted)
+        model = sio.load(fname, trusted=untrusted)
+        print(f"  Loaded: {fname}")
+        return model
     except Exception as e:
-        print(f"Error loading '{fname}': {e}")
-        sys.exit(1)
+        raise RuntimeError(f"Failed to load '{fname}': {type(e).__name__}: {e}") from e
 
 
+print("Loading preprocessor...")
 preprocessor = _load_skops_model(os.path.join(MODEL_DIR, "preprocessor.skops"))
 
+print("Loading ensemble models...")
 classification_model_data = {}
 for _o in CLASSIFICATION_OUTCOMES:
     _path = os.path.join(MODEL_DIR, f"ensemble_model_{_o}.skops")
     if os.path.exists(_path):
         classification_model_data[_o] = _load_skops_model(_path)
     else:
-        print(f"Warning: Model for {_o} not found at {_path}. Skipping.")
+        print(f"  Warning: Model for {_o} not found at {_path}. Skipping.")
 
 classification_models = {o: d["models"] for o, d in classification_model_data.items()}
 betas                 = {o: d["beta"]   for o, d in classification_model_data.items()}
@@ -119,7 +117,7 @@ for _o, _d in classification_model_data.items():
             _cal = _d["calibrator"]
         else:
             print(
-                f"Warning: outcome '{_o}' has calibrator_type='{_cal_type}'. "
+                f"  Warning: outcome '{_o}' has calibrator_type='{_cal_type}'. "
                 "Skipping non-isotonic calibrator (isotonic-only policy)."
             )
     elif "isotonic_calibrator" in _d and _d["isotonic_calibrator"] is not None:
@@ -138,11 +136,14 @@ ohe                     = preprocessor.named_transformers_["cat"]
 ohe_feature_names       = ohe.get_feature_names_out(CATEG_COLUMNS)
 processed_feature_names = np.concatenate([NUM_COLUMNS, ohe_feature_names])
 
+print(f"Models loaded: {list(classification_models.keys())}")
+
 
 # ---------------------------------------------------------------------------
 # SHAP background data
 # ---------------------------------------------------------------------------
 
+print("Building SHAP background...")
 np.random.seed(23)
 _n_background = 500
 
@@ -177,9 +178,10 @@ _background_data = {
                      ], _n_background),
 }
 
-_background_df = pd.DataFrame(_background_data)[FEATURE_NAMES]
-_X_background  = preprocessor.transform(_background_df)
+_background_df  = pd.DataFrame(_background_data)[FEATURE_NAMES]
+_X_background   = preprocessor.transform(_background_df)
 shap_background = shap.maskers.Independent(_X_background)
+print("SHAP background ready.")
 
 
 # ---------------------------------------------------------------------------
@@ -611,4 +613,7 @@ def icon_array(probability, outcome):
         plot_bgcolor="white",
         paper_bgcolor="white",
     )
-    return fig
+    return fig
+
+
+print("===== inference.py loaded successfully =====")