Re: [R] computationaly singular error!

2013-09-12 Thread Rose
I have 8 predictors and 1144 observations over 13 years. I think it is not the case. -- View this message in context: http://r.789695.n4.nabble.com/computationaly-singular-error-tp4675810p4675962.html Sent from the R help mailing list archive at Nabble.com.

Re: [R] computationaly singular error!

2013-09-12 Thread S Ellison
> Error in solve.default(H, g[!fixed]) : > system is computationally singular: reciprocal condition number = > 4.65795e-19 > Maybe you have more predictors than data? S Ellison *** This email and any attachments are confident

Re: [R] computationaly singular error!

2013-09-12 Thread Bert Gunter
... or some of the predictors are (near) linear combinations of others. This often occurs when some of the predictors are categorical/factors with lots of levels. Cheers, Bert On Thu, Sep 12, 2013 at 8:45 AM, S Ellison wrote: > > Error in solve.default(H, g[!fixed]) : > > system is computa

Re: [R] computationaly singular error!

2013-09-12 Thread Rose
Thanks Bert, all predictors are numerical. By the way, I have checked the collinearity of predictors by vif(). All vif of predictors were less than 2 that means none of the predictors are linear combinations of the others. Best, Rose -- View this message in context: http://r.789695.n4.nabble.

[R] computationaly singular error!

2013-09-10 Thread Rose
Hi there, I am trying to estimate the mlogit model. but I get the following error. Error in solve.default(H, g[!fixed]) : system is computationally singular: reciprocal condition number = 4.65795e-19 I have googled it and found out the collinearity between independent variables cause this err