Linear models with 1 and 2 explanatory variables
We initially described the simple normal linear model,
However when there are 2 explanatory variables, the corresponding formulae for the least squares estimates are unwieldy and they become impractically complex if there are more explanatory variables.
A different approach must be taken to describe the least squares estimates for a GLM with p explanatory variables. The next two pages will show that matrix notation provides a simple formula for the least squares estimates in a GLM.
In practice, it is not necessary to know formulae for the least squares estimates — you can use a computer to evaluate them.