* Augmenting A Design;
/*
Sometimes it is desirable to augment a design in an optimal way.
Suppose it is expected that a linear model will suffice for the
experiment and so the design in design1 is run.
*/
data design1;
input x1 x2;
cards;
-1 -1
1 -1
-1 1
0 0
0 0
;
run;
/*
Later, the experimenters realized that a second-order design
would be more appropriate. They constructed a sensible list of
design points to include (based on a face-centered CCD).
*/
data candidate1;
input x1 x2;
cards;
-1 -1
1 -1
-1 1
0 0
;
run;
data candidate2;
input y1 y2;
cards;
1 0
-1 0
0 1
0 -1
;
run;
/*
Then they evaluate the star points at a number of values, and find
the D-optimal augmentation at each.
*/
%macro augment_eg;
%do alpha=1 %to 4;
data candidate3;
set candidate2;
x1=&alpha*y1*.5;
x2=&alpha*y2*.5;
keep x1 x2;
run;
data candidates;
set candidate1 candidate3;
run;
proc optex data=candidates;
model x1|x2 x1*x1 x2*x2;
generate n=15 augment=design1;
examine design;
run;
%end;
quit;
%mend augment_eg;
%augment_eg;