Date:         Fri, 28 Mar 1997 19:08:13 -0500
Reply-To:     Pierre Duchesne <duchesne@DMS.UMONTREAL.CA>
Sender:       "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From:         Pierre Duchesne <duchesne@DMS.UMONTREAL.CA>
Subject:      Symbol problem
Content-Type: text/plain; charset=ISO-8859-1

Hello,
 
First, i ran PROC REG, and found fitted values from my regression.
Now i want to overlay my regression line with original observations.
Here is my program:
 
=======================================================
DATA cancer;
  INFILE 'cancer.dat';
  INPUT occupe $ fume mort;
  DROP occupe;
RUN;
 
PROC REG DATA=cancer;
  MODEL mort = fume / P;
  OUTPUT OUT=regout RESIDUAL=residu RSTUDENT=rstudent P=prevue;
RUN;
 
SYMBOL1 INTERPOL=;
SYMBOL2 INTERPOL=JOIN;
 
PROC GPLOT DATA=regout;
 PLOT mort*fume prevue*fume / OVERLAY;
 FILENAME gsasfile 'graph2.ps';
 GOPTIONS DEVICE=ps GACCESS=gsasfile;
RUN;
=======================================================
 
I tried everything:
SYMBOL1 INTERPOL=JOIN;
SYMBOL2 INTERPOL=;
 
or
SYMBOL1 INTERPOL=JOIN;
SYMBOL2 INTERPOL=NONE;
 
I tried also something like mort*fume=1 prevue*fume=2.
 
However, it doesn't work.  I can't assign different symbols statement
for my two plots.  With the preceding instructions, it is always the
same symbol statement who is used for the two plots.
 
I really want to fit my regression line by myself, because maybe i will
remove some points (outliers values), and plot a more robust regression
line on the data.  So i want to be able to join my own fitted values,
and have a scatter plot of my observations.  How can i do that?
 
Thank you very much for help!
Joyeuses Paques!
Pierre