ABSTRACT
This chapter examines linear models which are a widely used regression technique. It uses a generic data frame called mydata. It is built using an equation for a linear model. Linear models predict a continuous response variable in terms of one or more explanatory variables. Linear models are mathematical equations where the response variable can be predicted through the addition of the products of the explanatory variable and their associated slope. If practitioners are given a linear equation they can predict the value of a response variable from the explanatory variables simply by putting their values into the linear equation. To understand goodness-of-fit it is helpful to understand residuals and correlation. The measure of goodness-of-fit in Akaike's Information Criterion (AIC) is known as the log-likelihood which can be calculated from the residual sum of squares and sample size. If there are two models with the same number of parameters then one with the lowest log-likelihood will have the lowest AIC.
