Design a Robust Cross-Validation Strategy for Your Machine Learning Pipeline

You are an expert machine learning engineer specializing in model evaluation. Design a comprehensive cross-validation strategy for the following scenario: **Dataset Details:** - Dataset name/description: [DATASET_NAME] - Number of samples: [NUM_SAMPLES] - Number of features: [NUM_FEATURES] - Target variable type: [CLASSIFICATION/REGRESSION] - Class balance (if applicable): [BALANCED/IMBALANCED — specify ratio] - Presence of time-dependent data: [YES/NO] - Presence of grouped/hierarchical data: [YES/NO] **Model(s) being evaluated:** [MODEL_NAME(S)] **Please provide the following:** 1. **Recommended CV method:** Choose from k-fold, stratified k-fold, leave-one-out, time-series split, group k-fold, or nested cross-validation. Justify your choice based on the dataset characteristics. 2. **Optimal number of folds:** Recommend k value and explain the bias-variance tradeoff at this setting. 3. **Data leakage prevention:** Identify at least 3 potential sources of data leakage in the CV pipeline and how to prevent each. 4. **Preprocessing integration:** Explain how feature scaling, encoding, and feature selection should be handled within the CV loop. 5. **Evaluation metrics:** Recommend primary and secondary metrics appropriate for [CLASSIFICATION/REGRESSION] with justification. 6. **Code template:** Provide a Python code template using scikit-learn that implements the recommended CV strategy, including proper Pipeline usage. 7. **Statistical significance:** Suggest a method to determine whether performance differences between folds or models are statistically significant. 8. **Common pitfalls:** List 3 common mistakes practitioners make with this specific CV setup and how to avoid them. Format the response with clear headings, code blocks, and a summary decision table.