proc reg data=hartnagl;
model ftheft = fertil labor postsec mtheft/dw;
output out=hartreg p=p r=r; /* We import this to hartreg with p is prediction
r is the OLS residual*/
run;
quit;
Friday, October 19, 2007
Tuesday, October 16, 2007
Robust optimizaton
If our solution are subject to error(which is
not a constant), it is meaningful to consider
the so called Robust Optimization.
Comparing to global optimization, the robust
optimization may do not find the highest point
but the most stable peak.
(MIN MAX (WORST CASE)) is the robust solution.
not a constant), it is meaningful to consider
the so called Robust Optimization.
Comparing to global optimization, the robust
optimization may do not find the highest point
but the most stable peak.
(MIN MAX (WORST CASE)) is the robust solution.
Subscribe to:
Comments (Atom)