Hi I was wondering if anyone could help me in understanding the ARIMA notation for the input series. Basically i have a response time series y and I have x1 and x2 as my inputs. I did the prewhitening before calculating the crosscorrelation. The prewhitening filter for x1 is ARIMA(1,1,2) and for x2 is ARIMA(0,1,2). And based on the crosscorrelation, I'd like to include lag2 for x1 and lag3 for x2, but I'm not sure how to correctly code it in the "estimate" statement such that the previous whitening filter is taken into account. So far, i have:
proc arima data=fit
identify var=x1(1); estimate p=1 q=2 noint;
identify var=x2(1); estimate q=2 noint;
identify var=y crosscorr=(x1(1) x2(1));
estimate input=(??x1 ??x2) noint;
run;quit;
Thank you.