topic Re: Proc Gam in Statistical Procedures

Proc Gam

GS2 — Mon, 27 Nov 2023 19:09:35 GMT

Hello,

Is their a method to produce written 95% CI in proc gam? I can get the confidence bands on the displayed figure but would like the numbers as well. Thank you

ods graphics on;
proc gam data= fi.analysis plots=component(clm);
	model Impact_Factor = spline(FI_RFI) / dist=normal;
run; quit;
ods graphics off;

Re: Proc Gam

sbxkoenk — Mon, 27 Nov 2023 22:43:05 GMT

See the dataset work.SmoothingComponentPlot.

ods graphics on;
*ods trace on;
ods output SmoothingComponentPlot=work.SmoothingComponentPlot;
proc gam data= sashelp.shoes plots=component(clm);
	model Sales = spline(Inventory) / dist=normal;
run; quit;
ods graphics off;

Note there is also PROC GAMPL.

PROC GAMPL Contrasted with PROC GAM

https://go.documentation.sas.com/doc/en/statcdc/14.3/statug/statug_hpgam_overview02.htm

Or maybe you want to consider PROC LOESS?

Koen

Re: Proc Gam

Rick_SAS — Tue, 28 Nov 2023 13:56:45 GMT

Do not use PROC GAM. That is an old procedure that has been replaced by PROC GAMPL, which is more efficient and can scale to large data. In PROC GAMPL, you can use the

proc gampl data= sashelp.shoes plots=Components;
   model Sales = spline(Inventory) / dist=normal;
   output out=GamOut pred lower upper;
run;