I was attempting to use the residual resampling approach to generate 999
bootstrap samples of alpha and beta and find their confidence intervals.
However, I keep getting the error message:Error in nls(resample.mp ~
cases/(alpha + (beta * cases)), start = init.values, : singular
gradientafter R has only produced a few bootstraps.Could anyone suggest where I
am going wrong? Would greatly appreciate it!Mary
manpower<-read.table("http://www.creem.st-and.ac.uk/len/classes/mt3607/data/manhours_surgical.dat";,
header=TRUE)#attach dataattach(manpower)B<-999#number of data
pointsn<-dim(manpower)[1]#alpha level to use for the confidence
limitsalpha<-0.05#matrix that's going to contain the bootstrap
coefficientsboot.coef<-matrix(NA, B+1, 2)#fit the initial
modelinit.values<-c(alpha=20,beta=0)model<-nls(manhours~cases/(alpha+(beta*cases)),
start=init.values, trace=TRUE)pred<-predict(model)resid<-resid(model)#do the
bootstrapfor (i in 1:B){ #resample the residuals resample.ind!
ex<-sample(1:n,n,replace=T) resample.mp<-pred+resid[resample.index] #refit
the model init.values<-c(alpha=20,beta=0)
new.model<-nls(resample.mp~cases/(alpha+(beta*cases)),start=init.values,
trace=TRUE) #extract the parameter estimates
boot.coef[i,]<-coef(new.model)}#add the original parameter estimates
boot.coef[B+1,]<-coef(model)#calculate confidence intervalsfor(i in 1:2){
ci<-quantile(boot.coef[,i],probs=c(alpha/2,(1-alpha/2))) cat("residual
bootstrap confidence intervals for parameter",i,"are",ci,"\n")}
Miss your Messenger buddies when on-the-go? Get Messenger on your Mobile!
_________________________________________________________________
Win Indiana Jones prizes with Live Search
[[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.
