1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
|
---
title: "Miscellaneous examples"
date: "`r Sys.Date()`"
output: rmarkdown::html_vignette
vignette: >
%\VignetteIndexEntry{Miscellaneous examples}
%\VignetteEngine{knitr::rmarkdown}
\usepackage[utf8]{inputenc}
---
```{r echo=FALSE}
library(glmmTMB)
```
## Beta dispersion model
```{r simbeta1}
set.seed(1001)
N <- 1000
mean_pars <- c(1,2)
disp_pars <- c(1,2)
dd <- data.frame(x=rnorm(N))
m <- plogis(mean_pars[1]+mean_pars[2]*dd$x)
d <- exp(disp_pars[1]+disp_pars[2]*dd$x)
dd$y <- rbeta(N,shape1=m*d,shape2=(1-m)*d)
```
Fit models:
```{r modbeta1}
## location only
m1 <- glmmTMB(y~x,
family=beta_family(),
data=dd)
## add model for dispersion
m2 <- update(m1,dispformula=~x)
```
Fixed effects look close to theoretical values:
```{r coefbeta1}
fixef(m2)
```
AIC is insanely much better for the model with dispersion varying:
```{r AICbeta1}
bbmle::AICtab(m1,m2)
```
|
