simPop (version 2.1.3)

simInitSpatial: Generation of smaller regions given an existing spatial variable and a table.

Description

This function allows to manipulate an object of class simPopObj in a way that a new variable containing smaller regions within an already existing broader region is generated. The distribution of the smaller region within the broader region is respected.

Usage

simInitSpatial(
  simPopObj,
  additional,
  region,
  tspatialP = NULL,
  tspatialHH = NULL,
  eps = 0.05,
  maxIter = 100,
  nr_cpus = NULL,
  seed = 1,
  verbose = FALSE
)

Value

An object of class simPopObj with an additional variable in the synthetic population slot.

Arguments

simPopObj

an object of class simPopObj.

additional

a character vector of length one holding the variable name of the variable containing smaller geographical units. This variable name must be available as a column in input argument tspatial.

region

a character vector of length one holding the variable name of the broader region. This variable must be available in the input tspatial as well as in the sample and population slots of input simPopObj.

tspatialP

a data.frame (or data.table) containing three columns. The broader region (with the variable name being the same as in input region, the smaller geographical units (with the variable name being the same as in input additional) and a third column containing a numeric vector holding counts of persons. This argument or tspatialHH has to be provided.

tspatialHH

a data.frame (or data.table) containing three columns. The broader region (with the variable name being the same as in input region, the smaller geographical units (with the variable name being the same as in input additional) and a third column containing a numeric vector holding counts of households. This argument or tspatialP has to be provided.

eps

relative deviation of person counts if person and household counts are provided

maxIter

maximum number of iteration for adjustment if person and household counts are provided

nr_cpus

if specified, an integer number defining the number of cpus that should be used for parallel processing.

seed

optional; an integer value to be used as the seed of the random number generator, or an integer vector containing the state of the random number generator to be restored.

verbose

TRUE/FALSE if some information should be shown during the process

Author

Bernhard Meindl and Alexander Kowarik

Details

The distributional information must be contained in an input table that holds combinations of characteristics of the broader region and the smaller regions as well as population counts (which may be available from a census).

References

M. Templ, B. Meindl, A. Kowarik, A. Alfons, O. Dupriez (2017) Simulation of Synthetic Populations for Survey Data Considering Auxiliary Information. Journal of Statistical Survey, 79 (10), 1--38. tools:::Rd_expr_doi("10.18637/jss.v079.i10")

Examples

library(data.table)
data(eusilcS)
data(eusilcP)
library(data.table)

# no districts are available in the population, so we have to generate those
# we randomly assign districts within "region" in the eusilc population data
# each hh has the same district
simulate_districts <- function(inp) {
  hhid <- "hid"
  region <- "region"

  a <- inp[!duplicated(inp[,hhid]),c(hhid, region)]
  spl <- split(a, a[,region])
  regions <- unique(inp[,region])

  tmpres <- lapply(1:length(spl), function(x) {
    codes <- paste(x, 1:sample(3:9,1), sep="")
    spl[[x]]$district <- sample(codes, nrow(spl[[x]]), replace=TRUE)
    spl[[x]]
  })
  tmpres <- do.call("rbind", tmpres)
  tmpres <- tmpres[,-c(2)]
  out <- merge(inp, tmpres, by.x=c(hhid), by.y=hhid, all.x=TRUE)
  invisible(out)
}

eusilcP <- data.table(simulate_districts(eusilcP))
# we generate the input table using the broad region (variable 'region')
# and the districts, we have generated before.
#Generate table with household counts by district
tabHH <- eusilcP[!duplicated(hid),.(Freq=.N),by=.(db040=region,district)]
setkey(tabHH,db040,district)
#Generate table with person counts by district
tabP <- eusilcP[,.(Freq=.N),by=.(db040=region,district)]
setkey(tabP,db040,district)

# we generate a synthetic population
setnames(eusilcP,"region","db040")
setnames(eusilcP,"hid","db030")
inp <- specifyInput(data=eusilcP, hhid="db030", hhsize="hsize", strata="db040",population=TRUE)
if (FALSE) {
# use only HH counts
simPopObj <- simStructure(data=inp, method="direct", basicHHvars=c("age", "gender"))
simPopObj1 <- simInitSpatial(simPopObj, additional="district", region="db040", tspatialHH=tabHH,
tspatialP=NULL, nr_cpus=1)

# use only P counts
simPopObj <- simStructure(data=inp, method="direct", basicHHvars=c("age", "gender"))
simPopObj2 <- simInitSpatial(simPopObj, additional="district", region="db040", tspatialHH=NULL,
tspatialP=tabP, nr_cpus = 1)

# use P and HH counts
simPopObj <- simStructure(data=inp, method="direct", basicHHvars=c("age", "gender"))
simPopObj3 <- simInitSpatial(simPopObj, additional="district", region="db040", tspatialHH=tabHH,
tspatialP=tabP, nr_cpus = 1)
}