Control Parameters for Zero-inflated Count Data Regression
Various parameters that control fitting of zero-inflated regression models
using zeroinfl.
zeroinfl.control(method = "BFGS", maxit = 10000, trace = FALSE, EM = FALSE, start = NULL, ...)
method |
characters string specifying the |
maxit |
integer specifying the |
trace |
logical or integer controlling whether tracing information on
the progress of the optimization should be produced (passed to |
EM |
logical. Should starting values be estimated by the EM (expectation maximization) algorithm? See details. |
start |
an optional list with elements |
... |
arguments passed to |
All parameters in zeroinfl are estimated by maximum likelihood
using optim with control options set in zeroinfl.control.
Most arguments are passed on directly to optim, only trace is also
used within zeroinfl and EM/start control the choice
of starting values for calling optim.
Starting values can be supplied, estimated by the EM (expectation maximization)
algorithm, or by glm.fit (the default). Standard errors are
derived numerically using
the Hessian matrix returned by optim. To supply starting
values, start should be a list with elements "count" and "zero"
and potentially "theta" (for negative binomial components only) containing
the starting values for the coefficients of the corresponding component of the
model.
A list with the arguments specified.
Achim Zeileis <Achim.Zeileis@R-project.org>
## Not run:
data("bioChemists", package = "pscl")
## default start values
fm1 <- zeroinfl(art ~ ., data = bioChemists)
## use EM algorithm for start values
fm2 <- zeroinfl(art ~ ., data = bioChemists, EM = TRUE)
## user-supplied start values
fm3 <- zeroinfl(art ~ ., data = bioChemists,
start = list(count = c(0.7, -0.2, 0.1, -0.2, 0, 0), zero = -1.7))
## End(Not run)Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.