Generalized linear models (glm's) in Julia
To use the package, clone this repository and, in Julia, run
push(LOAD_PATH, "/path/to/repository/src/")
require("init.jl")The glmFit function in this package fits generalized linear models with the Iteratively Reweighted Least Squares (IRLS) algorithm. It is closer to the R function glm.fit than to R's glm in that the user is required to specify the model matrix as a matrixm, rather than in a formula/data specification.
A GlmResp object is created from the response vector, distribution and, optionally, the link. The available distributions and their canonical link functions are
Bernoulli (LogitLink)
Poisson (LogLink)
and the available links are
CauchitLink
CloglogLink
IdentityLink
InverseLink
LogitLink
LogLink
ProbitLink
The response in the example on p. 93 of Dobson (1990) would be written
rr = GlmResp(Poisson(), [18.,17,15,20,10,20,25,13,12])
At present the model matrix must be generated in the following awkward way
X = float64([1 0 0 0 0; 1 1 0 0 0; 1 0 1 0 0; 1 0 0 1 0; 1 1 0 1 0; 1 0 1 1 0; 1 0 0 0 1; 1 1 0 0 1; 1 0 1 0 1])
and the fit is
d = DensePred(X)
glmFit(d, rr)
d.beta
deviance(rr)