Forecasts with Unobserved Components Model

[jaweriahh]

I have a very simple question. I am running an Unobserved Components Model to decompose and forecast my data. I transform my data into log and then run the model. My code is given below

proc ucm data = metals;
id year interval = year;
model metal_indices=level1988;
level variance=0 noest;
slope plot=smooth;
cycle plot=smooth;
cycle plot=smooth;
estimate back=6;
forecast back=6 lead=6 plot=decomp;
run;

As I am using the log form of data after I obtain my forecasts and parameters I need to transform the data back. What code should I use for it?

[udo_sas · Accepted solution]

Hello -

As an example see "Predicting with model BSM4" chapter of this very nice paper called "the unobservable components model" by Prof. Tom Fomby of Department of Economics, Southern Methodist University available here: http://faculty.smu.edu/tfomby/eco5375/data/Unobservable%20Components%20Models/The%20Unobservable%20C... - a SAS program called BSM.sas that can be found in the appendix of this document.

Thanks,

Udo

[jaweriahh]

Thank you very much i used the code and it worked.