feat: add CalibratedClassifierCV and roc_curve imports for enhanced model evaluation

app/training/train_classifier.py

@@ -40,12 +40,14 @@ from sklearn.base import clone
 from sklearn.ensemble import GradientBoostingClassifier, RandomForestClassifier
 from sklearn.exceptions import ConvergenceWarning
 from sklearn.linear_model import LogisticRegression
+from sklearn.calibration import CalibratedClassifierCV
 from sklearn.metrics import (
     accuracy_score,
     f1_score,
     precision_score,
     recall_score,
     roc_auc_score,
+    roc_curve,
 )
 from sklearn.model_selection import StratifiedKFold, cross_val_predict, train_test_split
 from sklearn.neural_network import MLPClassifier
@@ -602,6 +604,13 @@ def _build_candidate_families() -> dict[str, list[Any]]:
     return families
 
 
+def _optimal_threshold(y_true: np.ndarray, y_prob: np.ndarray) -> float:
+    """Youden's J statistic: threshold that maximises sensitivity + specificity - 1."""
+    fpr, tpr, thresholds = roc_curve(y_true, y_prob)
+    j_scores = tpr - fpr
+    return float(thresholds[np.argmax(j_scores)])
+
+
 def _build_eval_pipeline(model: Any) -> Pipeline:
     return Pipeline(
         [