Sampling in statistics or machine learning is the process of selecting a subset of data from a larger dataset. Proper sampling ensures the subset is representative, reducing bias. It’s used in many contexts, from batch training to Monte Carlo methods.