Solved
Learner
Posts: 1

# Using Proc Model for forecasting

Hi,

I am trying to use proc model to run a time series regression yt = A*xt by 10 group ids and then use the regression result to forecast y(t+1) at x(t+1).  The coefficients came out with 10 groups. However the forecast result came out just as one group. I wonder how can I generate forecast by groups as well. My code is as follows.

data shock1; *generate x1 to xt then forecast y at x(t+1) and x(t+2)
r = 0; * 5 lagged values assumed be zero;
do i=1 to 5; output; end;
i=6; * impulses;
r = 1; output shock1;
r = .; * future values;
do i=7 to 8; output; end; run;

proc model data=regdata;
by groupID;
r = A1*lag5(1,r)+A2*lag5(2,r)+A3*lag5(3,r)+A4*lag5(4,r)+A5*lag5(5,r);

fit r/ sur outs=vcov (drop=_nused_) outest=est out=res;
solve / forecast data=shock1 out=prediction;

The out=prediction dataset only generate one group of forecast without groupID. I need the forecast to be also by groupID. I would appreciate it very much if anyone could help me on this.

Accepted Solutions
Solution
‎09-17-2016 09:37 AM
Super User
Posts: 19,769

## Re: Using Proc Model for forecasting

Should Shock1 have a groupID?

All Replies
Solution
‎09-17-2016 09:37 AM
Super User
Posts: 19,769

## Re: Using Proc Model for forecasting

Should Shock1 have a groupID?

Super User
Posts: 10,018

## Re: Using Proc Model for forecasting

```Plz post it at forecast forum ,since it is about SAS/ETS
```
☑ This topic is solved.