ljrf {ljr}R Documentation

Perform forward joinpoint selection algorithm with unlimited upper bound.

Description

This function performs the full forward joinpoint selection algorithm based on the likelihood ratio test statistic. The p-value is determined by a Monte Carlo method.

Usage

ljrf(y,n,tm,X,ofst,R=1000,alpha=.05)

Arguments

y the vector of Binomial responses.
n the vector of sizes for the Binomial random variables.
tm the vector of ordered observation times.
X a design matrix containing other covariates.
ofst a vector of known offsets for the logit of the response.
R number of Monte Carlo simulations.
alpha significance level of the test.

Details

The re-weighted log-likelihood is the log-likelihood divided by the largest component of n.

Value

pvals The estimates of the p-values via simulation.
Coef A table of coefficient estimates.
Joinpoints The estimates of the joinpoint, if it is significant.
wlik The maximum value of the re-weighted log-likelihood.

Author(s)

The authors are Michal Czajkowski, Ryan Gill, and Greg Rempala. The software is maintained by Ryan Gill rsgill01@louisville.edu.

References

Czajkowski, M., Gill, R. and Rempala, G. (2007). Model selection in logistic joinpoint regression with applications to analyzing cohort mortality patterns. To appear.

See Also

ljrk,ljrb,ljrf2

Examples

 N=20
 m=2
 k=0
 beta=c(0.1,0.1,-0.05)
 gamma=c(0.1,-0.05,0.05)
 ofst=runif(N,-2.5,-1.5)
 x1=round(runif(N,-0.5,9.5))
 x2=round(runif(N,-0.5,9.5))
 X=cbind(x1,x2)
 n=rep(10000,N)
 tm=c(1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10)
 eta=ofst+beta[1]+gamma[1]*tm
 if (m>0)
 for (i in 1:m)
  eta=eta+beta[i+1]*X[,i]
 if (k>0)
  for (i in 1:k)
   eta=eta+gamma[i+1]*pmax(tm-tau[i],0)
 y=rbinom(N,size=n,prob=exp(eta)/(1+exp(eta)))
 temp.ljr=ljrf(y,n,tm,X,ofst,R=1000)

[Package ljr version 1.1-0 Index]