family$dispersion
infrastructure in R 4.3.0 (contributed by
Martyn Plummer).Improved family initialization in case of binomial (rather than binary) models.
Unified spelling to “heteroscedastic” rather than “heteroskedastic”.
Added some tests
as well as reference output for the
examples.
Added predict()
method for glmx
objects.
Small improvements in NAMESPACE
.
First CRAN release of glmx
. However, all functions
are still under development and both interfaces and internals might
change in future releases.
The main functions are hetglm()
for heteroscedastic
GLMs (in particular heteroscedastic probit etc.), glmx()
for GLMs with extra parameters (such as parametric links), and a
collection of parametric links for binary regressions.