Machine Learning Model Evaluation and Selection Framework

You are a machine learning engineer and data scientist. I need to evaluate and select the best ML model for my problem. My task: [classification/regression/clustering/ranking]. Dataset: [size, features count, target variable]. Business constraint: [latency requirement, interpretability needs, class imbalance, available compute]. Guide me through: 1. Choosing the right evaluation metrics for my task (accuracy vs. F1 vs. AUC-ROC vs. RMSE — explain tradeoffs) 2. Setting up a proper train/validation/test split strategy 3. Cross-validation approaches: k-fold, stratified, time-series split 4. Baseline model to beat: simple heuristic or naive model 5. Model selection matrix: compare Logistic Regression, Random Forest, XGBoost, Neural Network 6. Hyperparameter tuning: Grid Search vs. Random Search vs. Bayesian Optimization 7. Handling overfitting: regularization, early stopping, ensemble methods 8. Model interpretability: SHAP values, feature importance, partial dependence plots 9. Production considerations: model size, inference speed, retraining strategy 10. Write Python code for a complete model evaluation pipeline with sklearn Provide the code for evaluating [classification/regression] with cross-validation.