library("countreg")
data("CrabSatellites", package = "countreg")
## default start values
fm1 <- nbreg(satellites ~ width + as.numeric(color), data = CrabSatellites)
## user-supplied start values
fm2 <- nbreg(satellites ~ width + as.numeric(color), data = CrabSatellites,
start = list(mu = c(0, 0, 0), theta = c(0.5)))Control Parameters for Negative Binomial Count Data Regression
Description
Various parameters that control fitting of negative binomial regression models using nbreg.
Usage
nbreg.control(method = "BFGS", maxit = 10000, start = NULL, hessian = TRUE,
dot = "separate", ...)
Arguments
method
|
characters string specifying the method argument passed to optim.
|
maxit
|
integer specifying the maxit argument (maximal number of iterations) passed to optim.
|
start
|
an optional list with elements “mu” and “theta” containing the coefficients for the corresponding component.
|
hessian
|
logical. Should the numerically approximated Hessian be computed to derive an estimate of the variance-covariance matrix? If FALSE and parameter hessA = FALSE in nbreg(), the variance-covariance matrix contains only NAs.
|
dot
|
character. Controls how two-part Formula’s are processed. See model.frame.Formula.
|
…
|
arguments passed to optim.
|
Details
All parameters in nbreg are estimated by maximum likelihood using optim with control options set in nbreg.control. Most arguments are passed on directly to optim and start controls the choice of starting values for calling optim.
Starting values can be supplied or are estimated by a Poisson regression in glm.fit (the default, starting values of coefficients in \(\theta\) are set to zero to ensure compatibility with NB1). Standard errors are derived using the analytical Hessian matrix or by numerical approximation of the Hessian.
Value
A list with the arguments specified.
See Also
nbreg