anova.DirichletRegModel: Compare Dirichlet Regression Models using an LRT

In DirichletReg: Dirichlet Regression

Description

This function allows for pairwise tests of Dirichlet regression models using a likelihood ratio test (LRT).

Usage

## S3 method for class 'DirichletRegModel'
anova(object, ..., sorted = FALSE)

Arguments

object	the model to be compared against those listed in ...
...	models to be tested against the one specified as object
sorted	should the models be sorted according to their numbers or parameters?

Details

The test statistic is computed LR=-2[log(La)-log(Lb)] where Li is the likelihood of model i with df equal to the difference of the number of parameters in the models.

Author(s)

Marco J. Maier

Examples

ALake <- ArcticLake
ALake$AL <- DR_data(ArcticLake[,1:3])
mod0 <- DirichReg(AL ~ 1, ALake)
mod1 <- DirichReg(AL ~ depth, ALake)
mod2 <- DirichReg(AL ~ depth + I(depth^2), ALake)
anova(mod1, mod0, mod2, sorted = TRUE)

Example output

Loading required package: Formula
Loading required package: rgl
Warning messages:
1: In rgl.init(initValue, onlyNULL) : RGL: unable to open X11 display
2: 'rgl_init' failed, running with rgl.useNULL = TRUE 
3: .onUnload failed in unloadNamespace() for 'rgl', details:
  call: fun(...)
  error: object 'rgl_quit' not found 
Warning in DR_data(ArcticLake[, 1:3]) :
  not all rows sum up to 1 => normalization forced
Analysis of Deviance Table

Model 3: DirichReg(formula = AL ~ depth + I(depth^2), data = ALake)
Model 1: DirichReg(formula = AL ~ depth, data = ALake)
Model 2: DirichReg(formula = AL ~ 1, data = ALake)

        Deviance N. par Difference df  Pr(>Chi)    
Model 3 -217.994      6                            
Model 1 -202.739      3     15.254  3  0.001612 ** 
Model 2  -79.059      9    138.935  3 < 2.2e-16 ***
---
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

DirichletReg documentation built on May 18, 2021, 5:06 p.m.