Using proc arima or proc forecast with defined coefficient

[Zuhdiyah]

I want to do some simulation on ARIMA model by changing the confidence level from 0.1 to 99.9. I have a defined coefficient or parameter estimation for each of parameters. How can I use PROC ARIMA or PROC FORECAST to do this? For example, I want to do simulation for ARIMA (4,0,1) and I already have coefficients for AR1-AR4 along with MA1. I just want to change my confidence level to generate different lower and upper value. Thank you.

[PaigeMiller]

In PROC ARIMA, both the FORECAST statement and the IDENTIFY statement has an option ALPHA= which allows you to change the alpha used for confidence intervals.

[Zuhdiyah]

yes, for that alpha I know. My question is how can I input my coefficient (parameter estimation) instead of SAS doing the estimation?

[PaigeMiller]

Well, I must say I was confused from your question whether you were asking about how to change alpha or how to do estimation for fixed model parameters, or both, or neither.

Please look at the NOEST option in the documentation for the ESTIMATE statement.

[Zuhdiyah]

sorry, if it was confusing.

Thanks for your suggestion, I will look into it.

[Zuhdiyah]

Can you please help me how to write the codes?

Let's say I have an ARIMA model (2,0,1) and the coefficients are as below:

AR1=0.5

AR2=0.8

MA1=3.5

I want to generate the forecast with confidence level for 0.5, 0.7, and 0.9, can the below codes work?

proc arima data=mydata;
identify var=hist; 
estimate ar=0.5 0.8 ma=3.5 noest; 
forecast id=date interval=month lead=12 alpha=0.5 out=myoutput;
run;

Thank you.

[PaigeMiller]

---

@Zuhdiyah wrote:

Can you please help me how to write the codes?

Let's say I have an ARIMA model (2,0,1) and the coefficients are as below:

AR1=0.5

AR2=0.8

MA1=3.5

I want to generate the forecast with confidence level for 0.5, 0.7, and 0.9, can the below codes work?

proc arima data=mydata;
identify var=hist; 
estimate ar=0.5 0.8 ma=3.5 noest; 
forecast id=date interval=month lead=12 alpha=0.5 out=myoutput;
run;

Thank you.

---

What is wrong with the code you show?

[Ksharp]

You can simulate ARIMA time series data by ARMASIM() in SAS/IML.
Check its documentation in support.sas.com

P.S.  @Rick_SAS might would write a blog about this topic .