12.2 Fitting a Regression Model

This section uses the techniques that are described in Chapter 4 to call the GLM procedure to perform a linear regression and to create graphs of the analysis in SAS/IML Studio. Techniques from Chapter 9 are used to add prediction bands and reference lines to the graphs in order to make the graphs more interpretable.

The regression model for this example uses variables from the Movies data set. As seen in previous chapters, there appears to be a relationship between the Budget of a movie and the US_Gross variable (see Figure 7.7). However, neither variable is normally distributed (see Figure 7.6), so Atkinson (1985) and others suggest applying a normalizing transformation to these variables. For these data, it makes sense to apply a logarithmic transformation to each variable (see Figure 8.5) and to model the log-transformed variables.