Page 1 of 1

Forecast evaluation

Posted: Tue Apr 05, 2011 6:19 pm
by startz
If you do a Forecast with only 1 observation in the forecast period you don't get a forecast evaluation. Perhaps some elements of the evaluation can't be computed, but others can and probably should be.

Re: Forecast evaluation

Posted: Thu Apr 07, 2011 12:11 pm
by EViews Glenn
My quick look suggests that the forecast evaluation is only not reported if there are no observations or if the sum-of-squared residuals is identically equal to zero (or there is an unrelated error). I'm not certain why the three reportable statistics are not shown (to be zero) in this case, but we'll look into it.

Re: Forecast evaluation

Posted: Thu Apr 07, 2011 12:25 pm
by startz
Try setting the forecast smpl to 7 7 in the attached.

Re: Forecast evaluation

Posted: Thu Apr 07, 2011 1:44 pm
by EViews Glenn
There are only 6 observation in the original workfile. Forecasting "7 7" means that we add an observation to the workfile with the dependent variable equal to NA. So no forecast evaluation summaries are possible.

Re: Forecast evaluation

Posted: Thu Apr 07, 2011 3:38 pm
by startz
Uh, DTD please.