*
* Example LUMBERAD.DAT from pp 558-562
*
open data lumberad.dat
data(format=free,org=columns) 1 100 ads sales
*
* Because the two series have such dramatically different scales and also vary
* quite a bit, we do an overlay graph with vertical ranges which don't overlap.
* (A simple overlay graph will be a tangle of lines).
*
graph(footer="Figure 13-15 Lumber Sales and Advertising",overlay=line,ovrange=.5) 2
# ads
# sales
*
@bjident ads
@bjident sales
*
boxjenk(const,ar=1) sales
@CrossCorr(report) ads sales
*
* Estimate the base transfer function model
*
boxjenk(const,inputs=1) sales
# ads 0 1 1
@regcorrs
*
* Add the noise model
*
boxjenk(const,ar=1,inputs=1,vcv) sales
# ads 0 1 1
@regcorrs
@crosscorr ads %resids