extrapolate: Add extrapolated values to demographic array.

In StatisticsNZ/dembase: Analysing Cross-Classified Data about Populations

Description

Given assumptions about the type and amount of change, extend a demographic array along a specified dimension.

Usage

extrapolate(
  object,
  along = NULL,
  labels,
  growth = 0,
  type = c("exponential", "linear", "missing")
)

## S4 method for signature 'DemographicArray'
extrapolate(
  object,
  along = NULL,
  labels,
  growth = 0,
  type = c("exponential", "linear", "missing")
)

Arguments

object	An object of class DemographicArray.
along	Name or index of dimension along which values are extrapolated. If along is omitted, extrapolate looks for a dimension with dimtype "time", "age", or "cohort" (in that order).
labels	Labels for the new elements to be added to dimension along.
growth	A single number, or a DemographicArray object specifying growth rates or increments.
type	The type of growth assumed: "exponential", "linear", or "missing". Can be abbreviated. Defaults to "exponential". See below for details.

Details

The along dimension must have dimscale "Intervals" or "Points", which means that it must have dimtype "age", "cohort", or "time". The intervals or points do not need to be evenly spaced.

Values may be extrapolated forward, from the end of the along dimension, or backward, from the start. See below for examples.

If growth is a demographic array, it will typically have been created by a call to growth. Use of a demographic array object allows growth rates or increments to differ within classifying variables: see growth for details.

If type is "exponential", then extrapolated values are generated using the formula y[n] = y[1] (1 + g)^(t[n] - t[1]), and if type is "linear", values are generated using y[n] = y[1] + (t[n] - t[1]) g, where g is growth. If type is "missing" the extra cells are filled with NAs.

Value

An object of the same class as object.

See Also

growth

Examples

## prepare data
library(datasets)
spend <- USPersonalExpenditure
names(dimnames(spend)) <- c("category", "year")
spend <- Values(spend)

## extrapolate assuming expenditure in all
## categories grows at 1% per annum
## (note that 'type' defaults to "exponential")
extrapolate(spend, labels = c(1965, 1970), growth = 0.01)

## linear increase of $1 per annum
extrapolate(spend, labels = c(1965, 1970), growth = 1, type = "linear")
extrapolate(spend, labels = c(1965, 1970), growth = 1, type = "l")

## grows at historial growth rate - same rate for all categories
gr <- growth(spend)
extrapolate(spend, labels = c(1965, 1970), growth = gr)

## grows at historical growth rates - rate specific to each category
gr <- growth(spend, within = "category")
extrapolate(spend, labels = c(1965, 1970), growth = gr)

## irregularly-spaced times
extrapolate(spend, labels = c(1972.25, 2001), growth = 0.01)

## backward extrapolation
extrapolate(spend, labels = c(1930, 1935), growth = 0.01)

## fill extra years with NAs
extrapolate(spend, labels = c(1965, 1970), type = "missing")


## example of extrapolation over age intervals
library(demdata)
deaths <- Values(VADeaths2)
gr <- growth(deaths, along = "age", within = c("sex", "residence"))
extrapolate(deaths, along = "age", labels = c("75-79", "80+"), growth = gr)

StatisticsNZ/dembase documentation built on Dec. 25, 2021, 4:49 p.m.