Edgard,
If your series is nonstationary, you will need to render it stationary
before forecasting. This means that you should run a dfgls or dfuller
test beforehand to be sure that the series you are forecasting is
stationary. Check the corrgram to know how many lags of autocorrelation
that you have to insert in the dfuller test if that is the test you will
use.
You may want to consider using the tsappend, add(#) where #= number
of periods over which you wish to forecast. This is done before
you use the predict. This will allow you to forecast over the proper
horizon(#).
If you are using the differenced series, you can always integrate
over it, if you know the proper starting value to get the raw series,
once you have estimated the forecast.
If you are using dx, you can obtain raw x from sum(dx), but the
proper starting value is crucial.
In its present form, the arima procedure in Stata cannot estimate or
predict nonstationary time series.
- Regards,
Bob Yaffee
Robert A. Yaffee, Ph.D.
Research Professor
Shirley M. Ehrenkranz
School of Social Work
New York University
home address:
Apt 19-W
2100 Linwood Ave.
Fort Lee, NJ
07024-3171
Phone: 201-242-3824
Fax: 201-242-3825
[email protected]
----- Original Message -----
From: Edgard Alfonso Polanco Aguilar <[email protected]>
Date: Monday, January 23, 2006 2:27 pm
Subject: st: problem with time series forecasting
> Dear statalisters,
>
> I've been working with time series in stata 9. I use the command
> "arima D.x, ar(p) ma(q)". Where D.x = x(t) - x(t-1). Then I need to
> forecast using that same model.
>
> The steps I use are:
> 1- Expand the number of observations to fill them with the
> forecasted values.
> 2- Use "predict" both static and dynamic.
>
> The problem I'm having is that predict gives me the values of x(t) -
> x(t-1) and when I use "predict xhat, y" stata gives me the forecast
> only one period ahead.
>
> Given this I have the following questions:
> 1- Is there any way to predict the real values (x instead of D.x)?
> 2- What does "dynamic" exactly does? I've been experimenting with that
> option and I find that the value of the forecasting depends on the
> value specified in the option.
>
> Thak you very much
> Edgard
>
> *
> * For searches and help try:
> * http://www.stata.com/support/faqs/res/findit.html
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
>
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/