The same as we did in correlation analysis, say we wish to look at the relationship between engine displacement (**displ**) and highway mpg (**hwy**); in other words, predicting **hwy** from **displ **using mpg data file.

proc reg data = mpg; model hwy = displ / stb; run; quit;

We see that the relationship between **hwy** and **displ** is positive (.19451) and based on the p-value (<.0001), we conclude this relationship is statistically significant. Hence, there is a statistically significant positive linear relationship between engine displacement and highway mpg.

SAS also provide several plots for diagnosing the model fit, which is very sweet.

From the two plot above, we conclude the model fit is OK except couple of outliers.

*See Also:*