This module covers decision theory, where we go beyond classification and minimize a loss function, meaning that we perform some optimization. We discuss different types of loss functions; the squared error emphasizes outliers more than the absolute error. As a numerical example, we look at a set of numbers and compute scalars that minimize (1) the sum of squared errors with respect to the set of numbers, (2) the sum of absolute errors. The latter approach is shown to be robust to outliers in the data.
