How to add explanatory variables in forecasting models?

[deleted_user]

Hi
Hope someone out there would help me.

I have 7 years historical data on hospital separations and want to forecast the future hospital demand for the next decade by selected variables such as category of disease, place of residence and place of treatment, using SAS/ETS.
A sample of data look like:

TimeID Separations Categories of disease, place of residence
01Jan2001 800,000 1 1
01Jan2002 900,000 1 1
.
.
01Jan2007 1,110,000 1 1

01Jan2001 900,000 1 2
01Jan2002 950,000 1 2
.
.
01Jan2007 1,120,000 1 2
01Jan2001 800,000 2 1
01Jan2002 900,000 2 1
.
.
01Jan2007 1,110,000 2 1

01Jan2001 900,000 2 2
01Jan2002 950,000 2 2
.
.
01Jan2007 1,120,000 2 2


The above is a simplified version of my data, just wanted to find out (i) how I can include the explanatory variables into forecasting models using Time Series Forecasting System (TSFS) and (ii) if I choose the 'fit the model automatically' option in the TSFS, does that mean the computer will then automatically select the best fit model for my data so that I don't have to test and compare each individual models such as using PROC ARIMA, PROC EXPAND, PROC VARMAX etc.

Thanks.

Paris

[sbb]

Suggest you check the SAS support website http://support.sas.com/ and the SEARCH FACILITY using the forum post subject to start. I found several references on the topic / subject posted.

Scott Barry
SBBWorks, Inc.

[deleted_user]

Sbb

I didn't find the answers from the website that you have directed me to. Can you please be more specific?

Paris

[sbb]

Using your SUBJECT, I used the website search facility and also Google advanced search with lowercase parameter site:sas.com to limit the search to the SAS.COM website, and found these two papers related to SAS/ETS software:

http://support.sas.com/rnd/app/papers/tsfs.pdf


Also, an acronym TSFS (Time Series Forecasting System) may help reveal additional reference materials.

The link below references a technical paper on PROC MDC, if useful:

http://support.sas.com/rnd/app/papers/mdc.pdf


Scott Barry
SBBWorks, Inc.

[deleted_user]

Thanks, sbb.

Paris