Cross Validation

Cross Validation Q&A

1What is cross-validation?
Answer: Repeated splitting for robust model performance estimation.
2What is k-fold CV?
Answer: Divide data into k folds; rotate validation fold k times.
3Why CV over single split?
Answer: Reduces variance of performance estimate.
4Stratified k-fold use?
Answer: Maintains class distribution in each fold for classification.
5Time-series CV?
Answer: Use forward chaining; never shuffle future into past.
6Nested CV purpose?
Answer: Unbiased estimate when tuning hyperparameters.
7Common k values?
Answer: 5 or 10 are common trade-offs of bias and compute.
8What is leakage in CV?
Answer: Preprocessing done before split leaks validation info into training.
9How avoid leakage?
Answer: Fit transforms inside each fold using pipelines.
10One-line summary?
Answer: Cross-validation provides more reliable model selection than one split.