DSsim (version 1.1.4)

make.population.description: Creates a Population.Description object

Description

Creates an object which describes a population. The values in this object will be used to create instances of the population

Usage

make.population.description(region.obj = make.region(),
  density.obj = make.density(), covariates = list(), N = numeric(0),
  fixed.N = TRUE)

Arguments

region.obj

the Region object in which this population exists (see make.region).

density.obj

the Density object describing the distribution of the individuals / clusters (see make.density).

covariates

Named list with one named entry per individual level covariate. Cluster sizes can be defined here. Each list entry should be another list with either one element or one element per strata allowing different population structures per strata. Each element of these lists should either be a data.frame containing 2 columns, the first the level (level) and the second the probability (prob). The cluster size entry in the list must be named 'size'. Alternatively the list element may be another list specifying the distribution in the first element and a named list in the second element with the distribution parameter.

N

the number of individuals / clusters in a population (1000 by default)

fixed.N

a logical value. If TRUE the population is generated from the value of N otherwise it is generated from the density description.

Value

object of class Population.Description

Details

#' The covariates argument should specify a list with one named element per covariate. If specifying the covariate values via a distribution this should be done in the form of a list. The first element should be one of the following: 'normal', 'poisson', 'ztruncpois' or 'lognormal'. The 'ztruncpois' distribution refers to a zero truncated Poisson distribution. The corresponding parameters that you must supply are detailed below. These should be added to a named list (each element named with the parameter name) containing the parameter values. See examples for implementation.

Distribution	Parameters
normal	mean	sd
poisson	lambda
ztruncpois	mean
lognormal	meanlog	sdlog

See Also

make.region, make.density, make.detectability

Examples

# NOT RUN {
# An example population can be created from the default values:
# - the default region
# - a constant density surface
# - and a population size of 1000
pop.desc <- make.population.description()

# To view an instance of this population
pop <- generate.population(pop.desc, make.detectability(), make.region())
plot(make.region())
plot(pop)

# An example population with covariates which vary by strata
# Make a multi strata region
poly1 <- data.frame(x = c(0,0,100,100,0), y = c(0,100,100,0,0))
poly2 <- data.frame(x = c(200,200,300,300,200), y = c(10,110,110,10,10))
coords <- list(list(poly1), list(poly2))
region <- make.region(coords = coords)
density <- make.density(region)

# Cluzter size is a zero truncated poisson with mean = 5 in strata 1 and a poisson with 
# lambda = 30 in strata 2.
covariate.list <- list()
covariate.list$size <- list(list("ztruncpois", list(mean = 5)),
                            list("poisson", list(lambda = 30)))
                            
# Animal height is generated from a lognormal distribution for both strata
covariate.list$height <- list(list("lognormal", list(meanlog = log(2), sdlog = log(1.25))))

# Animal sex is discrete/categorical, there are more females than males in strata 1 and equal
# numbers in strata 2
covariate.list$sex <- list(data.frame(level = c("male", "female"), prob = c(0.45,0.55)), 
                           data.frame(level = c("male", "female"), prob = c(0.5,0.5)))
                           
# Create covariate description
pop.desc <- make.population.description(region.obj = region, 
                                        density.obj = density, 
                                        covariates = covariate.list, 
                                        N = c(10,10))

# To view the covariate values
pop <- generate.population(pop.desc, detect = make.detectability(), region)
pop@population 
# Note that the covariate values have not affected the detectability (the scale parameter) to 
# do this we need to set the cov.param argument in make.detectability. See ?make.detectability
# }