topic Re: Logistic regression-calculate pred on out of time data in SAS Programming

Logistic regression-calculate pred on out of time data

Ronein — Sat, 14 Dec 2024 07:16:57 GMT

Hello friends ,

I have train data, test data, out of time data.

There are 3 things I want to do-

1-calculate model coefficients based on train data

2-calculate pred valued for each observation train+test data

3-calculate pred valued for each observation in out of time data

What is the way to perform task 3 using proc genmod ?

Please note that I tried to save the coefficients in a permanents data set (permanent library) but I cannot find this data set

proc genmod data=panel;
ods output parameterestimates=tbl_coefficients;/****Data set with Coeficients information that was calculated from train data only****/
weight weight;
model y=x / dist=binomial link=logit  type3 wald ;
output out=preds_tbl pred=P_BAD XBETA=logit;
store R_R.coefficients;
Run;


Data train(KEEP=X Y);
Do i=1 to 50;
x=ranuni(1234579);
p = 0.6;
u = rand("Uniform");
if (u < p) then Y=1;else Y=0;
output;
end;
Run;


Data test(KEEP=X Y);
Do i=1 to 25;
x=ranuni(1234579);
p = 0.6;
u = rand("Uniform");
if (u < p) then Y=1;else Y=0;
output;
end;
Run;

data panel;
set train(in=a) test(in=b);
if a then weight=1;else weight=0;/***Value 1 for train data, value 0 for test data**/
Run;


proc genmod data=panel;
ods output parameterestimates=tbl_coefficients;/****Data set with Coeficients information that was calculated from train data only****/
weight weight;
model y=x / dist=binomial link=logit  type3 wald ;
output out=preds_tbl pred=P_BAD XBETA=logit;
Run;



Data Out_of_time;
Do i=1 to 200;
x=ranuni(1234579);
p = 0.6;
u = rand("Uniform");
if (u < p) then Y=1;else Y=0;
output;
end;
Run;
/**I want to calculate pred on this data of out of time***/

Re: Logistic regression-calculate pred on out of time data

Ksharp — Sat, 14 Dec 2024 08:09:49 GMT

1-calculate model coefficients based on train data


Data train(KEEP=X Y);
Do i=1 to 50;
x=ranuni(1234579);
p = 0.6;
u = rand("Uniform");
if (u < p) then Y=1;else Y=0;
output;
end;
Run;


Data test(KEEP=X Y);
Do i=1 to 25;
x=ranuni(1234579);
p = 0.6;
u = rand("Uniform");
if (u < p) then Y=1;else Y=0;
output;
end;
Run;

data panel;
set train test;
Run;


proc genmod data=train;
ods output parameterestimates=tbl_coefficients;
model y=x / dist=binomial link=logit  type3 wald ;
 store PainModel ;
Run;

2-calculate pred valued for each observation train+test data


proc plm restore=PainModel;
   score data=panel out=Score_panel predicted LCLM UCLM / ilink; /* ILINK gives probabilities */
run;

3-calculate pred valued for each observation in out of time data


Data Out_of_time;
Do i=1 to 200;
x=ranuni(1234579);
p = 0.6;
u = rand("Uniform");
if (u < p) then Y=1;else Y=0;
output;
end;
Run;

proc plm restore=PainModel;
   score data=Out_of_time out=Score_Out_of_time predicted LCLM UCLM / ilink; /* ILINK gives probabilities */
run;

Re: Logistic regression-calculate pred on out of time data

Ronein — Sat, 14 Dec 2024 08:19:56 GMT

Thank you. Is it possible to save the coefficients in a permanent data set and then can use it in the future on any out of time data set to calculate predictions...??

Re: Logistic regression-calculate pred on out of time data

Ksharp — Sat, 14 Dec 2024 08:36:20 GMT

In my code:
ods output parameterestimates=tbl_coefficients;

already save these estimated parameters into dataset named 'tbl_coefficients' .

"can use it in the future on any out of time data set to calculate predictions...??"
Sure, of course. The following is an example for scoring dataset PANEL:

proc transpose data=tbl_coefficients out=tbl_coefficients2 prefix=_;
var Estimate;
id Parameter;
run;
data score_panel2;
if _n_=1 then set tbl_coefficients2;
set panel;
y_hat=logistic(_intercept+_x*x);
run;

Re: Logistic regression-calculate pred on out of time data

PaigeMiller — Sat, 14 Dec 2024 10:29:11 GMT

It's a lot easier if you use PROC LOGISTIC instead of PROC GENMOD. There doesn't seem to be a reason in your problem description that you would have to use GENMOD.

proc logistic data=train ... ;
    model y = x1 x2 ... ;
    score data=train out=trainpreds;
    score data=test out=testpreds;
    score data=outoftime out=outoftimepreds;
run;