Linear least squares

Linear least squares is the most popular statistical method to fit a model to some data. This fitting is performed in a very specific way: the square difference between the model and the data is minimized. This is usually called “solving the normal equations”, and this procedure gives the least squares method its name.

You will need to understand what the normal equations are and how they are a consequence of this minimization procedure. Finally this project will require that you solve a couple of example problems using linear regression, either by performing the solutions by hand or by writing a program that solves the normal equations in a simple case (this last option is more challenging, and also more rewarding).

The wikipedia page on Linear least squares contains a lot of information and links.