Simulate species mix data for model fitting.

species_mix.simulate(
  archetype_formula,
  species_formula,
  data,
  offset = NULL,
  nArchetypes = 3,
  alpha = NULL,
  beta = NULL,
  gamma = NULL,
  logTheta = NULL,
  powers = NULL,
  size = NULL,
  family = "bernoulli"
)

Arguments

archetype_formula: formula to simulate species_mix data, needs to have the format: cbind(spp1,spp2,spp3,...,sppN)~1 + x1 + x2
species_formula: formula to simulate species_mix species-specific responses, e.g: ~1
data: a matrix of variables to simulate data from.
offset: used to offset sampling effort for abundance data (log link function).
nArchetypes: number of groups to simulate.
alpha: coefficients for each species archetype. vector S long.
beta: coefficients for each species archetype. Matrix of G x number of parameters. Each row is a different species archetype.
gamma: coefficients for each species archetype. Matrix of S x number of parameters. Each row is a different species archetype.
logTheta: coefficients for the dispersion variables for negative.binomial and gaussian distributions - should be number of species long and on the natural log scale.
powers: Is the power parameter for Tweedie distribution.
size: Is for the binomial model and this represents the number of binomial trials per site, can be fixed or vary.
family: Which statistical distribution to simulate data for. 'bernoulli','binomial', 'gaussian', 'ippm', 'negative.binomial' and 'poisson'.

Examples

# \donttest{
archetype_formula <- stats::as.formula(paste0('cbind(',paste(paste0('spp',
1:20),collapse = ','),")~1+x1+x2"))
species_formula <- stats::as.formula(~1)
beta <- matrix(c(-3.6,0.5,
                 -0.9,1.0,
                  0.9,-2.9,
                  2.2,5.4),
                4,2,byrow=TRUE)
dat <- data.frame(y=rep(1,100),
                  x1=stats::runif(100,0,2.5),
                  x2=stats::rnorm(100,0,2.5))
simulated_data <- species_mix.simulate(archetype_formula,species_formula,
                                       data=dat, nArchetypes = 4, beta=beta,
                                       family="bernoulli")
#> Random alpha from normal (-1,0.5) distribution
# }