*
* SERIESF.DAT example from pp 257-258
*
open data seriesf.dat
calendar(m) 1995
data(format=free,org=columns) 1995:1 1998:12 demand
*
* Compute the trend rate, using STATISTICS to get the mean of the year over year
* difference of demand.
*
set d12 = demand-demand{12}
stats d12
compute at=%mean/12
*
set detrend = demand-at*t
diff(center) detrend
*
** The SPECTRUM procedure can be used to estimate the periodogram
** and the spectral density estimate.
*
@spectrum(periodogram,nologscale) detrend
@spectrum(window=tent,width=7,ordinates=48) detrend