panel {panel} | R Documentation |
The algorithm is discussed in Kalbfleisch and Lawless (1985) and further in Gentleman (1994).
panel(indata, qmatf, gamma, qderivf, npar, nstage, ncov, verbose = F, tol = 0.001)
indata |
indata is a list with one element for each individual. Each element of indata is a structure with 4 components. time: a vector containing the inspection times stage: a vector containing the state of the individual at the corresponding inspection time cov: a vector containing the covariate values at the inspection times len: the number of inspections on that individual |
npar |
npar is the number of parameters that are to be estimated |
nstage |
nstage is the number of stages/states in the model |
ncov |
ncov is the number of levels the covariate comes at |
qmatf |
qmatf is a function that accepts as input gamma (the parameters) and returns an array (ncov x nstage x nstage) with the ncov transition intensity matrices. |
gamma |
gamma is a vector (npar long) of estimates of the transition parameters. |
qderivf |
qderivf is a function that takes gamma as an argument and returns an array (npar x ncov x nstage x nstage) with the derivatives of the transition intensity matrices with respect to each of the parameters. |
verbose |
If verbose is set to TRUE then on each iteration the current estimate of theta and the current value of the score function are printed out. |
tol |
The tolerance used to determine convergence. |
panel returns a list with the parameter estimates, their standard errors and the log likelihood on convergence.
There are no known side effects.
Kalbfleisch J. D. and Lawless J. F. "The Analysis of Panel Data Under a Markov Assumption", JASA (1985), p. 863. Gentleman R. (1994). The Use of Covariate Information in Multi-state Markov Models. Technical Report 13, Department of Statistics, University of Auckland.
eddcmp, dclike.panel
#See the accompanying document "A Users Guide to panel" for #examples and code segments.