proc optmodel: saving solution status in a dataset

[Alain38]

Hello,

Is it possible to save in a dataset the solution status of proc optmodel?

I'm indeed running thousands of optimization problems such as:

%let byvar = grp;
proc optmodel printlevel=0;
	set OBS;
	set <str> ASSETS;
	set <num,str> GROUPS_ASSETS;
	set GROUPS;

	num grp{OBS};
	num MVC{GROUPS_ASSETS, ASSETS};
	num E{GROUPS_ASSETS};
	num target_var{GROUPS};

	read data MVC into ASSETS=[Name];
	read data MVC into OBS=[_N_] grp;
	read data MVC nomiss into GROUPS_ASSETS=[k=grp i=Name] {j in ASSETS} <MVC[k,i,j]=col(j)>;
	read data Predicted nomiss into [k=grp] {j in ASSETS} <E[k,j]=col(j)>;
	read data Target_Var into GROUPS=[grp] target_var;

	set BYSET = setof {i in OBS} &byvar.[i];
	num by;

	set OBS_BY = {i in OBS: &byvar.[i] = by};
	set Assets_BY = setof {o in OBS_BY, <(grp[o]),a> in GROUPS_ASSETS} a;
	var W{Assets} >= 0;

	maximize Expected = sum{i in Assets_BY}W[i]*E[by,i];
	con sum{i in Assets_BY, j in Assets_BY}W[i]*MVC[by,i,j]*W[j] = target_var[by];
	con BUDGET: sum{i in Assets_BY}W[i] = 1;

	num W_sol {GROUPS_ASSETS};
	do by = BYSET;
		put by=;
		solve with NLP / feastol = 1E-7 logfreq = 0 maxiter = 10000;
		for {i in Assets_BY} W_sol[by,i] = W[i].sol;
	end;

	create data Weights_NoShort from [&byvar i] W=W_sol;
quit;

However, I noticed that in a few cases, the NOTE in the log is not "Optimal" but:

  - NOTE: Maximum number of iterations reached

  - NOTE: Infeasible

Saving them in a dataset would allow me to quickly check which by group did not solve correctly.

Thank you in advance,

[RobPratt · Accepted solution]

Yes, you can use the _solution_status_ parameter:

num W_sol {GROUPS_ASSETS};
str solstatus {BYSET};
do by = BYSET;
   put by=;
   solve with NLP / feastol = 1E-7 logfreq = 0 maxiter = 10000;
   for {i in Assets_BY} W_sol[by,i] = W[i].sol;
   solstatus[by] = _solution_status_;
end;

create data Weights_NoShort from [&byvar i] W=W_sol;
create data solstatusdata from [&byvar] solstatus;
 

[Alain38]

It's working perfectly, thank you!