Dear list,
I'm currently analyzing some count data using a hurdle model. I've used
the rcspline.eval function in the Hmisc-library to contruct the spline
terms for the regression model, and what I want in the end is the ability
to compute coefficients and confidence intervals for different changes in
the smooth function as well as plotting the smooth function along with the
confidence interval at the values of the x-variable.
An example using the "zero"-part of the hurdle model:
library(Hmisc)
library(pscl)
# Simulate some data
set.seed(1)
y<-c(rep(0,50),rnbinom(50,0.9,0.2))
x<-sin(y)+rnorm(100)
# Set up the spline terms
ssp<-rcspline.eval(x,inclx=T)
# Fit the model and construct the smooth function
f<-hurdle(y~ssp)
knots<-attr(ssp,"knots")
coef<-f$coefficients$zero
w<-rcspline.restate(knots,coef)
fun<-eval(attr(w,"function"))
The coefficient for a change in x from -0.1 to 0.1 is fun(0.1)-fun(-0.1).
My question is therefore how do I compute the confidence interval for this
change?
This is easy to do with the Design-library for the "zero"-part but as far
as I know, zero-truncated negative binomial data can't be fitted using
Design's functions as they are. Does someone know any neat tricks that
would make this possible?
Any help on this would be greatly appreciated. Thanks for your time.
/Erik Lampa, Statistician, Uppsala University Hospital, Sweden
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org 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.
