Hi everyone,
I tried to understand the specifications of the syntax CROSSCORR of PROC ARIMA for an interrupted time series analysis. Can someone explain explicitly the meaning of the numbers put in the parentheses?
My dataset contains 100 rows. I created a variable INTERVENTION=1 for the start of the intervention, transformed the data with natural logarithm, and number the time variable order with date_id.
data dataset;
input year 1-4 month 6-7 number 9-13 date_id 15-16 intervention 18 log_number 20-25;
datalines;
...
2013 10 79064 88 0 11.278
2013 11 45969 89 0 10.735
2013 12 29014 90 0 10.275
2014 1 21057 91 1 9.954
2014 2 12326 92 1 9.419
2014 3 8417 93 1 9.038
...
;
run;
I want to run a ARIMA(1,0,1)(0,1,0)12 with a seasonality and test for the effect of the intervention. Do I need to put date_id or month in crosscorr
a)
proc arima data=dataset;
identify var=log_number(12) nlag=20 crosscorr=intervention;
estimate p=(1) q=(1) input=( /(1)intervention);
run;
b)
proc arima data=dataset;
identify var=log_number(12) nlag=20 crosscorr=intervention(1);
estimate p=(1) q=(1) input=( /(1)intervention);
run;
b)
proc arima data=dataset;
identify var=log_number(12) nlag=20 crosscorr=intervention(12);
estimate p=(1) q=(1) input=( /(1)intervention);
run;
d)
proc arima data=dataset;
identify var=log_number(12) nlag=20 crosscorr=(intervention date_id);
estimate p=(1) q=(1) input=( /(1)intervention, date_id);
run;
Many thanks,