[R] Determine the order of Time series
Richard M. Heiberger
rmh at temple.edu
Tue Jul 15 17:49:57 CEST 2014
Yes, of course.
Please look at the tsdiagplot function in the HH package.
install.packages("HH") ## if you don't have it yet
The example shows 0<p<3 and 0<q<3
You can change that to 6 and 10 with
X.loop <- arma.loop(X, order=c(6,0,10))
## The example dataset gives warnings and trapped errors because these
values are too big for this dataset.
X.dal <- diag.arma.loop(X.loop, x=X)
X.diag <- rearrange.diag.arma.loop(X.dal)
X.diagplot <- tsdiagplot(armas=X.loop, ts.diag=X.diag, lwd=1)
## You will need a big plotting window to see this, the screen isn't big enough
pdf("example.pdf", width=24, height=12)
This is my first example with two-digit q, and I discovered that the
ar:p,ma:q values are sequenced
alphabetically instead of numerically. That shows up as the
ar:0,ma:10 column appearing
between the ar:0,ma:1 and ar:0,ma:2 columns. I will fix it in the
next release of HH (probably late August).
Please let me (and the list) know how this works for you.
On Mon, Jul 14, 2014 at 3:49 PM, 张天添 <pennytianzh at hotmail.com> wrote:
> Dear R help team,
> I am currently stuck on how to determine the order of a data. I have plotted the ACF and PACF of the data and it does not seem to cuts off. Is that possible to plot AIC BIC of some order(p,q) say, 0<p<6 and 0<q<10 on the same graph so that I can compare the value of AIC and BIC to determine the best model to fit? (Please see attached picture that can clear describe what I am looking for.)
> Or, is there any other way to determine the order of the data?
> Thank you so much for your time
> Kind Regrads
> R-help at r-project.org mailing list
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
More information about the R-help