# Singularity issues when fitting an ARIMAX model

When I want to apply an ARIMA model like this:

arimax(dataMat, order=c(7,0,0), xreg=xreg)


it raises an error

"singular and could not compute QR decomposition of Hessian" for xreg.

How can I fit proper model by using xreg values?

-
supply your data. There could be a number of reasons, why you get this error. Multicolinearity would be prime suspect. –  mpiktas Jan 12 '12 at 13:42
The xreg data is a 913*99 matrix. 99 columns defines boolean values for each row. So, each cell value is either 0 or 1. –  Ahmet Altun Jan 12 '12 at 14:34
check that those boolean values do not sum to the constant. Or easier do qr(cbind(1,xreg))\$rank and look at the result. If it is less than 100 (99+1), you have multicolinearity issues, which can be solved by removing columns from xreg. –  mpiktas Jan 12 '12 at 14:47