VarReg: VarReg: Semi-parametric mean and variance regression
In VarReg: Semi-Parametric Variance Regression

VarReg

R Documentation

VarReg: Semi-parametric mean and variance regression

Description

Methods for fitting semi-parametric mean and variance models, with normal or censored data. Also extended to allow a regression in the location, scale and shape parameters.

Details

This package provides functions to fit semi-parametric mean and variance regression models. These models are based upon EM-type algorithms, which can have more stable convergence properties than other algorithms for additive variance regression models.

The primary function to use for linear and semi-parametric mean and variance models is semiVarReg. This function also is able to fit models to censored outcome data. There is also a plot function for these models called plotVarReg. A search function has also been produced in order to assist users to find the optimal number of knots in the model (searchVarReg).

The other functions that are of particular use are lssVarReg and its plot function plotlssVarReg. This uses the skew-normal distribution and combines the EM algorithm with a coordinate-ascent type algorithm in order to fit a regression model in the location, scale and shape, therefore extending the semi-parametric models to non-normal data.

Multivariate models can be fit with semiVarReg.multi and lssVarReg.multi