options nodate nonumber ps=55 ls=80;

title 'RPD -- Chapter 7 -- Variable Selection: Linthurst Data';

data linthurst (drop=obsnum);
infile 'linthurst.dat';
input obsnum loc $ type $ biomass salinity pH K Na Zn;
run;


symbol value=dot;

proc reg;
model biomass = salinity pH K Na Zn / selection=rsquare;
run;

plot rsq.*in. / nomodel nostat;
run;


model biomass = salinity pH K Na Zn /
   selection=rsquare adjrsq mse cp aic sbc best=1;
run;


plot mse.*in. / nomodel nostat;
run;


plot adjrsq.*in. / nomodel nostat;
run;


plot cp.*np. / nomodel nostat cmallows=blue;
run;


plot aic.*in. / nomodel nostat;
run;


plot sbc.*in. / nomodel nostat;
run;


model biomass = salinity pH K Na Zn /
   selection=forward sle=0.50;
run;


model biomass = salinity pH K Na Zn /
   selection=backward sls=0.10;
run;


model biomass = salinity pH K Na Zn /
   selection=stepwise sle=0.50 sls=0.15;
run;


model biomass = pH Na / p;   /* option p prints PRESS as a side effect */
run;