The ARIMA Procedure |
Using ARIMA Procedure Statements |
The IDENTIFY, ESTIMATE, and FORECAST statements are related in a hierarchy. An IDENTIFY statement brings in a time series to be modeled; several ESTIMATE statements can follow to estimate different ARIMA models for the series; for each model estimated, several FORECAST statements can be used. Thus, a FORECAST statement must be preceded at some point by an ESTIMATE statement, and an ESTIMATE statement must be preceded at some point by an IDENTIFY statement. Additional IDENTIFY statements can be used to switch to modeling a different response series or to change the degree of differencing used.
The ARIMA procedure can be used interactively in the sense that all ARIMA procedure statements can be executed any number of times without reinvoking PROC ARIMA. You can execute ARIMA procedure statements singly or in groups by following the single statement or group of statements with a RUN statement. The output for each statement or group of statements is produced when the RUN statement is entered.
A RUN statement does not terminate the PROC ARIMA step but tells the procedure to execute the statements given so far. You can end PROC ARIMA by submitting a QUIT statement, a DATA step, another PROC step, or an ENDSAS statement.
The example in the preceding section illustrates the interactive use of ARIMA procedure statements. The complete PROC ARIMA program for that example is as follows:
proc arima data=test; identify var=sales nlag=24; run; identify var=sales(1); run; estimate p=1; run; estimate p=1 q=1; run; outlier; run; forecast lead=12 interval=month id=date out=results; run; quit;
Copyright © 2008 by SAS Institute Inc., Cary, NC, USA. All rights reserved.