Builds a leakage-safe tabular classification pipeline end to end: a clean train/test split, preprocessing inside a Pipeline/ColumnTransformer, baseline vs. tuned models with cross-validated hyperparameter search, and honest metrics (accuracy, precision/recall/F1, ROC-AUC, confusion matrix) on a held
Builds a leakage-safe tabular classification pipeline end to end: a clean train/test split, preprocessing inside a Pipeline/ColumnTransformer, baseline vs. tuned models with cross-validated hyperparameter search, and honest metrics (accuracy, precision/recall/F1, ROC-AUC, confusion matrix) on a held-out set. Use whenever the user wants to "train a model", "predict a class/label", "build a classifi