Dear S-users,
In mid-April there was a question posted about the employment of
bootstrapping after model fitting. There were many great replies from Vito,
Prof. Ripley and others. Unfortunately, one question still remained
unanswered.
For example, I am fitting a GAM using the following input:
fit<-gam(y~x1+lo(x2)+lo(x3)+x4, family=quasi(log,mu), na.action=na.omit)
where,
x1 is a factor
x2-x4 are continuous
I would like to run a bootstrap on the coefficients from the fit and tried
using:
a<-bootstrap(data.frame,coef(glm(fit)),B=1000,trace=F)
followed by,
summary(a)
(Vito's suggestion)
After the 10 minute process was complete I received the following message:
Warning messages:
z0 is outside range (-.25, .25), this indicates extreme bias, assumptions
underlying BCa interval may be violated in: limits.bca(object, probs,
frame.eval.jack = frame.eval.jack)
Does anyone know why this would occur? Or, how I can rectify the problem? I
know that others have received this same message and apologize if the
answer has already been posted. Any help would be appreciated and I thank
you in advance for your time.
Daniel
""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
Daniel.G.C. Rainham
Climate and Health Research Program
Department of Earth and Atmospheric Sciences
University of Alberta
Office: 3-13 Tory Bldg
Fax: 780.492.7598
drainham@ualberta.ca
"The lottery is a tax on people who flunked math." --Monique Lloyd
""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news
|