topic Forecast missing values Proc ESM in SAS Forecasting and Econometrics

Forecast missing values Proc ESM

pilicasalv — Tue, 31 Oct 2017 16:36:08 GMT

Hi folks

I am new to SAS and I am contemplating to use the command PROC ESM. I have 3 monthly employment series from Jan 2000 to July 2016. The 3 series have some missing values. Also 2 of the variables don't have any values at the beginning. Is this command the right one? Should I do something different if the first values of the series are missing?

Another option is to use moving average to estimate the missing months. What command should I use in SAS?

Thanks

Maria

Re: Forecast missing values Proc ESM

Astounding — Tue, 31 Oct 2017 16:33:00 GMT

I hope somebody has a better way ... I always have had to jump through multiple hoops when there are missing values at the beginning of the series. My steps:

(1) Forecast missing values, using something like ID month_begin_date interval=monthly; That still leaves missing values at the beginning of the series.

(2) Create an artificial variable, with month beginning dates, but it backwards order. So the first MONTH_BEGIN_DATE has the highest value for this artificial variable, and the last MONTH_BEGIN_DATE has the lowest value. But the artificial variable must still take on values that represent the first day of the month. Sort the data by this artificial variable.

(3) Forecast missing values, using something like ID ARTIFICIAL_DATE interval=monthly;

(4) Drop the ARTIFICIAL_DATE.

That's a lot to ask if you are new to SAS. So I hope somebody out there has a better way.

Re: Forecast missing values Proc ESM

pilicasalv — Tue, 31 Oct 2017 16:34:01 GMT

Thank you! Will take into account your suggestion 🙂