[R] Multivariate Autoregressive Model calibration and residual testing

Zebouni, Stephane (Exchange) SZebouni at bear.com
Thu Mar 9 11:40:01 CET 2006



I am using the mAr package to calibrate an Multivariate model (size 3,
order 12). I am trying to do the two following things:


1.	I would like to calibrate the model using not a single time
series, but several of them: each time series should be seen as one
"independent" realisation of the mAr process; for instance this happens
when you have a time series with lacking data (holes) : if I have two
time series and a hiatus in the middle, I would like my obective
function in the calibration to be the sum of the two classical objective
function, one for each continuous time series. Is there a function in R
that could do this type of estimation? Or is there a nice way to do that
by modifying the objective function within the mAr.est function (I tried
but the code is too opaque..)?


2.	Second, I would like to use the multivariate Ljung-Box test for
the residuals of this multivariate AR, but I couldn't find it - (I found
just the univariate case) - does it exist somewhere?


Thanks a lot for your help!!



-------------- next part --------------

Bear Stearns is not responsible for any recommendation, solicitation, 
offer or agreement or any information about any transaction, customer 
account or account activity contained in this communication.

More information about the R-help mailing list