[R-pkgs] New package: ordinal

Rune Haubo rhbc at imm.dtu.dk
Tue Mar 16 11:45:14 CET 2010


This is to announce the new R-package ‘ordinal’ that implements
cumulative link (mixed) models for ordinal (ordered categorical) data
(http://www.cran.r-project.org/package=ordinal/).

The main features are:
-	scale (multiplicative) as well as location (additive) effects
-	nominal effects for a subset of the predictors (denoted partial
proportional odds when the link is the logistic)
-	structured thresholds, e.g. assuming symmetry or equidistant thresholds
-	random effects via the Laplace approximation and adaptive
Gauss-Hermite quadrature in the location-part of the model.
-	a range of standard link functions
-	flexible link functions where an extra link function-parameter
bridges the log-log, probit and c-loglog links (log-gamma), and
cloglog and logistic links (Aranda-Ordaz)
-	a suite of optimizers including an efficient Newton scheme.
-	works for binomial observations (a special case of ordinal data).
-	a suite of methods including anova, addterm, dropterm, profile,
confint, plot.profile, predict, in addition to the standard print and
summary methods.
-	an important special case is the proportional odds model (with
random effects).
-	a range of examples illustrates how to use the functions.

Future additions will include:
-	more general random effect structures: multiple (crossed and nested)
and vector-valued random effects.
-	profile methods for variance parameters in mixed effect models.
-	helpful package vignettes.
-	implementation of core functions in C.

Comments, critique, suggestions, wishes and contributions are always
highly appreciated.

Kind regards
Rune

-- 
Rune Haubo Bojesen Christensen

PhD student, M.Sc. Eng.
Phone: (+45) 45 25 33 63
Mail: rhbc at imm.dtu.dk

DTU Informatics, Section for Statistics
Technical University of Denmark, Build. 305, Room 122,
DK-2800 Kgs. Lyngby, Denmark



More information about the R-packages mailing list