In dataspekt/crodi: Computation of Croatian regional development index

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

Computation of regional development index

Computation of regional development index using package crodi has two basic steps:

• Normalization of indicators using function normalize().
• Computation of composite index using function devindex().

This example illustrates two methods which were used for computation of Croatian regional development index in years 2017 and 2013, some other variants and convenience functions.

First, load functions and data provided by this package:

library("crodi")
data(crodi2017)
data(crodi2013)

Method from 2017

Normalization of indicator values

Normalization of indicators is done by standardization of original indicator values to z-score values with mean of 100 and standard deviation of 10. Direction of values is reversed where appropriate.

crodi2017$sinc <- normalize(crodi2017$inc)
crodi2017$srev <- normalize(crodi2017$rev)
crodi2017$semp <- normalize(crodi2017$emp, reverse = TRUE)
crodi2017$sedu <- normalize(crodi2017$edu)
crodi2017$spop <- normalize(crodi2017$pop)
crodi2017$sagi <- normalize(crodi2017$agi, reverse = TRUE)

Calculation of penalised composite index

Computation of composite index is based on Mazziotta-Pareto method which features penalization of index values dependent on variability of indicator values. Function devindex() applies penalization by default.

crodi2017$di.calc <- devindex(~ sinc + srev + semp + sedu + spop + sagi, crodi2017)
head(crodi2017[c("di","di.calc")])

Method from 2013

Normalization of indicator values

In this method, normalization of indicators requires reference values. These are provided (for year 2013) in this package.

data(refval2013)

Normalization of indicator values is done using modified Min-Max method. To use it, set method = "mms" and specify appropriate reference value for each indicator. Direction of values is reversed where appropriate.

crodi2013$sinc <- normalize(crodi2013$inc, method = "mms", refval = refval2013$inc)
crodi2013$srev <- normalize(crodi2013$rev, method = "mms", refval = refval2013$rev)
crodi2013$semp <- normalize(crodi2013$emp, method = "mms", refval = refval2013$emp, reverse = TRUE)
crodi2013$sedu <- normalize(crodi2013$edu, method = "mms", refval = refval2013$edu)
crodi2013$spop <- normalize(crodi2013$pop, method = "mms", refval = refval2013$pop)

Calculation of weighted composite index

To calculate weighted composite index specify indicator weights as a named list.

indicator.weights <- list(sinc = .25, srev = .15, semp = .3, sedu = .15, spop =.15)

These indicator weights are input to devindex() function. Argument penalize is set to FALSE as penalization is not used and argument aggregate is set to "sum" as weighted indicator values are summed, not averaged.

crodi2013$di.calc <- devindex(~ sinc + srev + semp + sedu + spop, crodi2013, penalize = FALSE, wts = indicator.weights, aggregate = "sum")
head(crodi2013[c("di","di.calc")])

Please note that in year 2013 regional development index for a specific group of local administrative units was further corrected, which is not covered here.

Other variants

By setting both wts and penalize = TRUE, weighted composite index that is further penalized can be calculated. Note that in the case of weighted indicators penalty depends on unbalance in weighted values.

crodi2013$di.calc.pen <- devindex(~ sinc + srev + semp + sedu + spop, crodi2013, penalize = TRUE, wts = indicator.weights, aggregate = "sum")
head(crodi2013[c("di","di.calc","di.calc.pen")])

Penalization

Function penalize() will output penalty as defined in Mazziotta-Pareto method. Input is a data frame with normalized indicators.

pen <- penalize(crodi2017[c("sinc", "srev", "semp", "sedu", "spop", "sagi")])
head(pen)

dataspekt/crodi documentation built on May 30, 2019, 4:06 a.m.