Chapter Contents
Chapter Contents
Previous
Previous
Next
Next
The REG Procedure

Example 55.11: Plotting Variance Inflation Factors

This example uses the REG procedure to create plots from a data set. The variance inflation factors (output by the OUTVIF option in the previous example) are plotted against the ridge regression control values k. The following statements create Output 55.11.1:

   data b (keep=_RIDGE_ x1-x3 x1x2 x1x1);
      set b;
      if _TYPE_='RIDGEVIF';
      label x1='variance inflation factor';
   run;

   legend3 position=(top right inside) across=3 
           cborder=black cframe=white
           label=(color=blue position=(top center)
                  'independent variables')
           value=('X1' 'X2' 'X3' 'X1X2' 'X1X1');
   symbol1 c=blue   /*v=circle  */;
   symbol2 c=yellow /*v=x       */;
   symbol3 c=cyan   /*v=triangle*/;
   symbol4 c=green  /*v=square  */;
   symbol5 c=orange /*v=diamond */;
   title 'Variance Inflation Factors of Acetylene Data';

   proc reg data=b;
      var _RIDGE_ x3 x1x2 x1x1;
      model x1=x2 / noprint;
      plot (x1 x2 x3 x1x2 x1x1)*_RIDGE_
           / nomodel nostat legend=legend3 overlay
             vaxis = 0 to 75 by 25 cframe=ligr
             haxis = 0 to .02 by .002;
      footnote "Note: the VIF at k=0 is 7682 for X1, " 
               "6643 for X1X1, 345 for X1X2, and 320 for X2";
   run;

The GPLOT procedure can create the same plot with the following statements. The resulting display is not shown in this report.

   axis1 label=(a=90 r=0 'variance inflation factor')
         order=(0 to 75 by 25) minor=none;
   proc gplot data=b;
      plot (x1 x2 x3 x1x2 x1x1)*_RIDGE_
           / legend=legend3 overlay frame
             vaxis = axis1
             haxis = 0 to .02 by .002 hminor=0;
      footnote "Note: the VIF at k=0 is 7682 for X1, " 
               "6643 for X1X1, 345 for X1X2, and 320 for X2";
   run;

Output 55.11.1: Using PROC REG to Plot the VIFs
regx2b.gif (5861 bytes)

Chapter Contents
Chapter Contents
Previous
Previous
Next
Next
Top
Top

Copyright © 1999 by SAS Institute Inc., Cary, NC, USA. All rights reserved.