BookmarkSubscribeRSS Feed
LNA2021
Calcite | Level 5
data ARIMA_ASTHMA_AB;
Set ASTHMA_ab;
step= (Quarter>='2020-Q1');
ramp= time_after;
if Quarter < '2020-Q1' then ramp=0;
Run;
Trial (1);
proc arima data=ARIMA_ASTHMA_AB;
identify var= Preval_qtr_ResprtyAnti (2,4) crosscorr=(step(2,4) ramp(2,4));
estimate p=2 q=(4) input=(step ramp) method=ml outmodel=AB_AS3;
run; quit;
Trial(2);
proc arima data=ARIMA_ASTHMA_AB;
identify var= Preval_qtr_ResprtyAnti (2,4) crosscorr=(step ramp);
estimate p=1 q=(4) input=(step ramp) method=ml outmodel=AB_AS5;
run; quit;

Hello,

I am using ITS (ARIMA) to study the impact of intervention on drug utilzation.

Based on the above codes: My question is when to use crosscorr= [Step(1,4) Ramp(1,4)] and when to use only crosscorr= [Step Ramp]?

I know in order to choose between different ARIMA models, I need to check SBC and AIC (the lower the better model).

Also, I attached the output for both trial 1 and 3 SAS codes. Could you please advise which model is better and why?

Thanks in advance.

1 REPLY 1
sbxkoenk
SAS Super FREQ

Hello @LNA2021 ,

 

I have moved your question to the

SAS Forecasting and Econometrics board (under the Analytics header).

You will get much better answers here.

 

Note there's a lot of literature on this subject.
And many approaches are taken : ARIMA(X) , State Space Models (SSM), Mixed models , ...

But to re-assure you : PROC ARIMA can definitely be used for interrupted time series analysis.

 

Cheers,

Koen

sas-innovate-2024.png

 

Time is running out to save with the early bird rate. Register by Friday, March 1 for just $695 - $100 off the standard rate.

 

Check out the agenda and get ready for a jam-packed event featuring workshops, super demos, breakout sessions, roundtables, inspiring keynotes and incredible networking events. 

 

Register now!

Multiple Linear Regression in SAS

Learn how to run multiple linear regression models with and without interactions, presented by SAS user Alex Chaplin.

Find more tutorials on the SAS Users YouTube channel.

Discussion stats
  • 1 reply
  • 255 views
  • 0 likes
  • 2 in conversation