rpql: Regularized PQL for Joint Selection in GLMMs

Share:

Performs joint selection in Generalized Linear Mixed Models (GLMMs) using penalized likelihood methods. Specifically, the Penalized Quasi-Likelihood (PQL) is used as a loss function, and penalties are then "added on" to perform simultaneous fixed and random effects selection. Regularized PQL avoids the need for integration (or approximations such as the Laplace's method) during the estimation process, and so the full solution path for model selection can be constructed relatively quickly.

Author
Francis K.C. Hui, Samuel Mueller, A.H. Welsh
Date of publication
2016-10-07 09:23:08
Maintainer
Francis Hui <fhui28@gmail.com>
License
GPL-2
Version
0.5

View on CRAN

Man pages

build.start.fit
Constructs a start fit for use in the 'rpql' function
calc.marglogL
Calculate the marginal log-likelihood for a GLMM fitted using...
gendat.glmm
Simulates datasets based on a Generalized Linear Mixed Model...
lseq
Generates a sequence of tuning parameters on the log scale
nb2
A negative binomial family
rpql
Joint effects selection in GLMMs using regularized PQL.
rpql-package
Joint effects selection in GLMMs using regularized PQL
rpqlseq
Wrapper function for joint effects selection in GLMMs using...
summary.rpql
Summary of GLMM fitted using regularized PQL.

Files in this package

rpql
rpql/NAMESPACE
rpql/R
rpql/R/rpql-main.R
rpql/R/auxilaryfunctions.R
rpql/MD5
rpql/DESCRIPTION
rpql/man
rpql/man/lseq.Rd
rpql/man/rpql-package.Rd
rpql/man/nb2.Rd
rpql/man/build.start.fit.Rd
rpql/man/rpqlseq.Rd
rpql/man/calc.marglogL.Rd
rpql/man/rpql.Rd
rpql/man/gendat.glmm.Rd
rpql/man/summary.rpql.Rd