Modified ARMA function

Date: Sun, 04 Aug 2002 23:40:04 -0700
From: krishna kumar <kriskumar at mailandnews.com>
Subject: [R] Modified ARMA function

 R-guRus ,

ARMA function in tseries,  seems to be calculating the AR coeff 's as

 coef <- lm(xx[,1]~xx[,lag$ar+1])$coef    [*snipped* from around  line
77,]

This is only to get preliminary estimates of the AR (and later also of the MA) coefficients. The final estimates are computed by conditional least squares (ie. minimizing function err defined in arma).

I'd like to  modify this model with another term somewhat in these lines

lm(xx[,1] ~xx[,lag$ar+1]+mvgsignal)$coef

where mvgsignal is a moving average signal based on some indicators, the
question
is could i simply hack into tseries and do this and hope all is well ,

is there a cleaner
way

There is a cleaner way: Use the function arima from ts and use xreg=mvsignal (if mvsignal is exogenous).

of specifying arbitrary parameters (additions) to GARCH and other
estimators?.

Please enlighten.

thanks in advance,

Krishna

best
Adrian

--
Dr. Adrian Trapletti             Phone :             +41 (0) 1 994 5631
Trapletti Statistical Computing  Mobile:             +41 (0)76 370 5631
Wildsbergstrasse 31              Fax   :             +41 (0) 1 994 5633
CH-8610 Uster                    Email :  mailto:a.trapletti at bluewin.ch
Switzerland                      WWW   : http://trapletti.homelinux.com