generate_data: Generate dummy data with several distributions

In inbo/inlatools: Diagnostic Tools for INLA Models

Generate dummy data with several distributions

Description

All distributions share the same latent variable η_{ij} = a + b_i with b_i = N(0, σ_r)

Usage

generate_data(
  a = 0,
  sigma_random = 0.5,
  n_random = 20,
  n_replicate = 10,
  nb_size = 1,
  b_size = 5,
  zero_inflation = 0.5
)

Arguments

a	the intercept of the latent variable
sigma_random	The standard error for the random effect σ_r
n_random	the number of random effect levels (groups)
n_replicate	the number of observation per random effect level
nb_size	the size parameter of the negative binomial distribution. Passed to the size parameter of NegBinomial
b_size	the size parameter of the binomial distribution. Passed to the size parameter of \link[stats]{Binomial}
zero_inflation	the probability the the observed value stems for the a point mass in zero

Details

• The Poisson distribution uses λ = e^{η_{ij}}

• The negation binomial distribution uses μ = e^{η_{ij}}

• The binomial distribution uses π_{ij} = e^{η_{ij}}/(e^{η_{ij}}+ 1)

Value

A data.frame

• ìd the id of the random effect

• eta the latent variable

• zero_inflation use the point mass in zero

• poisson the Poisson distributed variable

• zipoisson the zero-inflated Poisson distributed variable

• negbin the negative binomial distributed variable

• zinegbin the zero-inflated negative binomial distributed variable

• binom the binomial distributed variable

See Also

Other utils: plot.dispersion_check(), plot.distribution_check()

Examples

set.seed(20181202)
head(generate_data())

inbo/inlatools documentation built on Sept. 17, 2022, 2:13 p.m.