Calculates the log-likelihood function of a GLM.
Currently only the gaussian and the bernoulli family
are implemented.
Usage
glm.ll(mu, y, phi=1, family="gaussian", k=1)
Arguments
mu
n x 1, predicted regression function
y
n x 1, responses
phi
scalar, nuisance parameter (sigma^2 for the gaussian and
inverse gaussian families, nu for the gamma family)
family
text string, family of distributions (e.g.
"gaussian" or "bernoulli", see details below)
k
integer > 0, parameter for the negative binomial
Value
log-likelihood value
Details
Implemented are the "gaussian" family
(with links "identity" and "log"), the "bernoulli" family
(with links "logit" and "probit"), the "gamma" family
(with link "inverse"), the "poisson" family (with link "log"),
the "inverse.gaussian" family (with link "inverse.squared") and
the "negative.binomial" (with its canonical "log" type link).
The default value k=1 leads to the geometric distribution (as a
special case of the negative binomial).