I motivate, derive, and discuss the bias-variance decomposition, which forms the theoretical basis of model validation in machine learning.