extra_fams: Family functions for Student's-t, Beta, Zero-Inflated and...

In GLMMadaptive: Generalized Linear Mixed Models using Adaptive Gaussian Quadrature

Family functions for Student's-t, Beta, Zero-Inflated and Hurdle Poisson and Negative Binomial, Hurdle Log-Normal, Hurdle Beta, Gamma, and Censored Normal Mixed Models

Description

Specifies the information required to fit a Beta, zero-inflated and hurdle Poisson, zero-inflated and hurdle Negative Binomial, a hurdle normal and a hurdle Beta mixed-effects model, using mixed_model().

Usage

students.t(df = stop("'df' must be specified"), link = "identity")
beta.fam()
zi.poisson()
zi.binomial()
zi.negative.binomial()
hurdle.poisson()
hurdle.negative.binomial()
hurdle.lognormal()
hurdle.beta.fam()
unit.lindley()
beta.binomial(link = "logit")
Gamma.fam()
censored.normal()

Arguments

link	name of the link function.
df	the degrees of freedom of the Student's t distribution.

Note

Currently only the log-link is implemented for the Poisson, negative binomial and Gamma models, the logit link for the beta and hurdle beta models and the identity link for the log-normal model.

Examples

# simulate some data from a negative binomial model
set.seed(102)
dd <- expand.grid(f1 = factor(1:3), f2 = LETTERS[1:2], g = 1:30, rep = 1:15,
                  KEEP.OUT.ATTRS = FALSE)
mu <- 5*(-4 + with(dd, as.integer(f1) + 4 * as.numeric(f2)))
dd$y <- rnbinom(nrow(dd), mu = mu, size = 0.5)

# Fit a zero-inflated Poisson model, with only fixed effects in the 
# zero-inflated part
fm1 <- mixed_model(fixed = y ~ f1 * f2, random = ~ 1 | g, data = dd, 
                  family = zi.poisson(), zi_fixed = ~ 1)

summary(fm1)


# We extend the previous model allowing also for a random intercept in the
# zero-inflated part
fm2 <- mixed_model(fixed = y ~ f1 * f2, random = ~ 1 | g, data = dd, 
                  family = zi.poisson(), zi_fixed = ~ 1, zi_random = ~ 1 | g)

# We do a likelihood ratio test between the two models
anova(fm1, fm2)

#############################################################################
#############################################################################

# The same as above but with a negative binomial model
gm1 <- mixed_model(fixed = y ~ f1 * f2, random = ~ 1 | g, data = dd, 
                  family = zi.negative.binomial(), zi_fixed = ~ 1)

summary(gm1)

# We do a likelihood ratio test between the Poisson and negative binomial models
anova(fm1, gm1)

GLMMadaptive documentation built on Oct. 17, 2023, 9:07 a.m.