I am using arima to develop a time series regression model, I am using arima b/c I have autocorrelated errors. Several of my independent variables are categorical and I have coded them as factors . When I run ARIMA I don't get any warning or error message, but I do not seem to get estimates for all the levels of the factor. Can/how does ARIMA handle factors in xreg? here is some example code: sim.ar <- arima.sim(list(ar = c(0.4, 0.4)), n = 1000) z<- factor(rep((1:5),200)) arima(sim.ar,order=c(2,0,0),xreg=z) I only get a single estiamte for xreg. Am I thinking about this wrong (I expected 4). thank you, Spencer [[alternative HTML version deleted]]
Try this: set.seed(1) sim.ar <- arima.sim(list(ar = c(0.4, 0.4)), n = 1000) z<- gl(5, 1, 1000) zm <- model.matrix(~z)[,-1] arima(sim.ar, order=c(2,0,0), xreg = zm) On 1/15/07, sj <ssj1364 at gmail.com> wrote:> I am using arima to develop a time series regression model, I am using arima > b/c I have autocorrelated errors. Several of my independent variables are > categorical and I have coded them as factors . When I run ARIMA I don't > get any warning or error message, but I do not seem to get estimates for all > the levels of the factor. Can/how does ARIMA handle factors in xreg? > > > here is some example code: > > sim.ar <- arima.sim(list(ar = c(0.4, 0.4)), n = 1000) > z<- factor(rep((1:5),200)) > > arima(sim.ar,order=c(2,0,0),xreg=z) > > > I only get a single estiamte for xreg. Am I thinking about this wrong (I > expected 4). > > thank you, > > > Spencer > > [[alternative HTML version deleted]] > > ______________________________________________ > R-help at stat.math.ethz.ch mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. >
>From the help pagexreg: Optionally, a vector or matrix of external regressors, which must have the same number of rows as 'x'. Note, not a factor. (It handles your factor like the integer vector it is internally.) If you want a design matrix you need to make one, something like include.mean = FALSE, xreg = model.matrix(~z) On Mon, 15 Jan 2007, sj wrote:> I am using arima to develop a time series regression model, I am using arima > b/c I have autocorrelated errors. Several of my independent variables are > categorical and I have coded them as factors . When I run ARIMA I don't > get any warning or error message, but I do not seem to get estimates for all > the levels of the factor. Can/how does ARIMA handle factors in xreg? > > > here is some example code: > > sim.ar <- arima.sim(list(ar = c(0.4, 0.4)), n = 1000) > z<- factor(rep((1:5),200)) > > arima(sim.ar,order=c(2,0,0),xreg=z) > > > I only get a single estiamte for xreg. Am I thinking about this wrong (I > expected 4). > > thank you, > > > Spencer > > [[alternative HTML version deleted]] > > ______________________________________________ > R-help at stat.math.ethz.ch mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. >-- Brian D. Ripley, ripley at stats.ox.ac.uk Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ University of Oxford, Tel: +44 1865 272861 (self) 1 South Parks Road, +44 1865 272866 (PA) Oxford OX1 3TG, UK Fax: +44 1865 272595