UNDER DEVELOPMENT

A R package that fit regression mixture model - group-based trajectory modeling (GBTM).

Longitudinal studies are often employed on several disciplines like finance, econometrics, psychology, sociology, biology, ...

The aims of this package is to give tools to work with this situation. On the one hand by divide data in some clusters and on the other hand by fitting longitudinal trajectories in each clusters.

Package Functionality

TrajeR support several distributions

• Censored (or regular) Normal distribution ;
• Zero Inflated Poisson distribution ;
• Poisson ditribution ;
• Bernouilli distribution ;
• Beta distribution (only withe Likelihood) ;
• Non linear regression.

The trajectories of each clusters are modeling by polynomials. We have the possibility to use Non Linear models too.

Installation

To get the current released version from CRAN:

# Comming soon

To get the current development version from Github:

## install dev version of trajeR from github
devtools::install_github("gitedric/trajeR")

## load trajeR package
library(trajeR)

Usage

The main function of the package is

trajeR(Y, A, X = NULL, TCOV = NULL, ng, degre, degre.nu = 0, 
       Model, Method = "L", 
       ssigma = FALSE, ymax = max(Y) + 1, ymin = min(Y) - 1,
       hessian = TRUE, itermax = 100, paraminit = NULL, 
       EMIRLS = TRUE, refgr = 1,
       fct = NULL, diffct = NULL, nbvar = NULL, nls.lmiter)

For more details about the usage of this package you can read the vignette documentation (coming soon).

Contact

Cédric NOEL

cedric.noel@univ-lorraine.fr