PROC UCM: do not produce prognosis (even if lead=1)

[asolk]

Hi, 

I am running a forecast, for at least 1 month long. I see the following note in the SAS:
NOTE: Number of non-missing predictor values in the forecast horizon is less than the value in the LEAD= option of the FORECAST statement. The value in the LEAD= option is reduced to 0.

So, I do not have any forecast data in the output 😞

I use the following code:

proc ucm data=all_TOTAL1;
   model all_50 = VA_all;
   id termin interval=month;
   irregular;
   level var=0 noest;
   randomreg batProgram / plot=smooth;
   season length=12 var=0 noest plot=smooth;
   estimate plot=(panel residual);
   forecast plot=forecasts lead=1;
run;

I have 31 non-missing variables (there is no missing in the datafile at all), data per month.

• All_50 (dependent) is the number of sold products,
• VA_all (predictor) is the number of days per month when the company conduct any promotion campaigns on the products.
• batProgram is used as a flag: 0 if no campaign and 1 if any campaign during the month

Forecast was fine before I have added a predictor variable VA_all (when I have run forecast based on the number of sold products only). But now I have these data and would like to integrate them into the analysis.

What is wrong in the code? Or data? Maybe 31 rows is quite small for this type of analysis?

Thanks in advance!

[sbxkoenk · Accepted solution]

Hello,

You need future values for your independent variables!

Do you have future values for your independent (=predictor) variables?

Cheers,

Koen

[asolk]

It works perfectly 🙂 Thank you so much, Koen!