Introduction
A projection finds the closest point in a subspace to a given vector — like the shadow of a stick cast straight down onto the ground. Projections are how we compress data (PCA), fit models (least squares) and remove unwanted components. We start with the simplest case: projecting onto a line through the origin.