topic Re: Question related to ARIMA modeling in SAS Forecasting and Econometrics

Question related to ARIMA modeling

munitech4u — Wed, 03 Apr 2013 13:30:31 GMT

Hi,

For ARIMA modeling, I have got the (P,D,Q,) as (1,1,1). But I am getting confused to form the equation, I have the below coefficients:

Parameter	MU	MA1,1	AR1,1
MA1,1	0.002	1.000	0.962
AR1,1	0.000	0.962	1.000
MU	1.000	0.002	0.000

I have the following 3 obs forecasted after differentiation:

Obs	Forecast	Std Error	95% Confidence Limits
95	-0.7976	6.6431	-13.8179	12.2226
96	0.6003	6.7904	-12.7086	13.9092
97	0.0578	6.8123	-13.2940	13.4097

Re: Question related to ARIMA modeling

Snurre_SAS — Wed, 03 Apr 2013 13:56:04 GMT

Hi,

I'm not entirely sure what you want to achieve but the first part of your output is not the actual parameter estimates. Instead, I think, you have shown the correlations of the parameter estimates. The actual parameter estimates can be located elsewhere in the output.

If you want to be able to write out the estimated equation you might find some input in this example: http://support.sas.com/documentation/cdl/en/etsug/65545/HTML/default/viewer.htm#etsug_arima_gettingstarted08.htm

At the bottom is an example where an ARIMA(1,1,1) is estimated.

Thanks,

Snurre

Re: Question related to ARIMA modeling

munitech4u — Wed, 03 Apr 2013 15:35:49 GMT

Thank you for the link. That was helpful. Though I have question: if I have to substitute the values of B and a(t) at a given time t, what would be there value.? And once we form this equation do we need to integrate, before forecasting?

Re: Question related to ARIMA modeling

Snurre_SAS — Wed, 03 Apr 2013 18:37:08 GMT

This link describes the general notation used when working with ARIMA models: http://support.sas.com/documentation/cdl/en/etsug/65545/HTML/default/viewer.htm#etsug_arima_gettingstarted13.htm.

Trying to setup the model outside of PROC ARIMA - for instance in Excel - is very tricky to do and not something I would recommend. It is not the same calculations as you would have when working for instance with simple linear regression. Forecasts are calculated as either infinite memory or finite memory forecasts which is based on the type of estimation method. More information can be found here: http://support.sas.com/documentation/cdl/en/etsug/65545/HTML/default/viewer.htm#etsug_arima_details32.htm

Thanks,

Snurre

Re: Question related to ARIMA modeling

munitech4u — Wed, 03 Apr 2013 18:44:31 GMT

Thnx!!

Re: Question related to ARIMA modeling

munitech4u — Wed, 03 Apr 2013 18:52:52 GMT

Which value will be assigned to random error a(t)?