BookmarkSubscribeRSS Feed
Jack_Smitherson
Fluorite | Level 6

Hi everyone, 

 

Proc logistic has a "unit" statement option, to attain an estimate on the association of interest by certain increments.

 

Can the same be done using proc arima? For example: 

 

proc arima data=data;
identify var=&ylist crosscorr=(&xlist);
estimate input=(&xlist) p=1 method=ml plot;
run;
forecast printall;
run;

quit;

 

Note: &xlist = continuous exposure/independent variable. 

1 REPLY 1
rselukar
SAS Employee

There is no "UNIT" statement in ARIMA.  I suspect you want to find how the response variable changes as the values of some input variable change.  Kind of scoring or what-if analysis.  In ARIMA there is no simple way to do it.  You could manually provide the desired input variable values in the forecast region (in the input data set) and reforecast.