fix: update calibration and precision-recall plotting functions to load data from files

app/training/generate_paper_figures.py

@@ -530,11 +530,14 @@ def plot_score_distribution(ml: dict) -> None:
 def plot_calibration(ml: dict) -> None:
     best = ml.get("_best_model", "LightGBM")
     data = ml.get(best, {})
-    if not data or "y_true" not in data:
+    y_prob_path = MODELS / "lgbm_cv_probs.npy"
+    y_true_path = MODELS / "lgbm_cv_ytrue.npy"
+    if not y_prob_path.exists():
+        print("  SKIP  No calibration data")
         return
 
-    y_true = np.array(data["y_true"])
-    y_prob = np.array(data["y_prob"])
+    y_true = np.load(y_true_path)
+    y_prob = np.load(y_prob_path)
 
     from sklearn.metrics import brier_score_loss
     brier = brier_score_loss(y_true, y_prob)
@@ -568,12 +571,14 @@ def plot_calibration(ml: dict) -> None:
 
 def plot_precision_recall(ml: dict) -> None:
     best = ml.get("_best_model", "LightGBM")
-    data = ml.get(best, {})
-    if not data or "y_true" not in data:
+    y_prob_path = MODELS / "lgbm_cv_probs.npy"
+    y_true_path = MODELS / "lgbm_cv_ytrue.npy"
+    if not y_prob_path.exists():
+        print("  SKIP  No precision-recall data")
         return
 
-    y_true = np.array(data["y_true"])
-    y_prob = np.array(data["y_prob"])
+    y_true = np.load(y_true_path)
+    y_prob = np.load(y_prob_path)
     prec, rec, _ = precision_recall_curve(y_true, y_prob)
     ap = average_precision_score(y_true, y_prob)
     baseline = y_true.mean()