Movatterモバイル変換

COINr

Full documentation is available atCOINr’swebsite

COINr is a high-level R package which offers a fully-flexibledevelopment and analysis environment for composite indicators andscoreboards. It is inspired by, and broadly follows the methodology oftheJRC/OECDHandbook on Composite Indicators and theCompetenceCentre for Composite Indicators and Scoreboards, which alsosupported the development of the package (seeAcknowledgements).

The main features can be summarised as features forbuilding, features foranalysis and features forvisualisation and presentation.

Building features:

• Flexible and fast development of composite indicators with no limitson aggregation levels, numbers of indicators, highly flexible set ofmethodological choices.
• Denomination by other indicators
• Screening units by data requirements
• Imputation of missing data, by a variety of methods
• Data treatment using Winsorisation and nonlineartransformations
• Normalisation (scaling) using a variety of methods
• Weighting using either manual weighting, PCA weights orcorrelation-optimised weights.
• Aggregation of indicators using a variety of methods which can bedifferent for each aggregation level.

Analysis features:

• Detailed indicator statistics, and data availability withinaggregation groups
• Multivariate analysis, including quick functions for PCA, and adetailed correlation analysis and visualisation
• Easy “what if” analysis - very quickly checking the effects ofadding and removing indicators, changing weights, methodologicalvariations
• Full global uncertainty and sensitivity analysis which can check theimpacts of uncertainties in weighting and many methodologicalchoices

Visualisation and presentation:

• Statistical plots of indicators - histograms, violin plots, dotplots, scatter plots and more
• Bar charts, stacked bar charts and tables for presenting indicatordata and making comparisons between units
• Correlation plots for visualising correlations between indicatorsand between aggregation levels

COINr also allows fast import from theCOINTool and fast export to Excel.

Installation

COINr is on CRAN and can be installed by running:

# Install released version from CRANinstall.packages("COINr")

The development version, which may be slightly more up-to-date, canbe installed from GitHub:

# Install development version from GitHubdevtools::install_github("bluefoxr/COINr")

This should directly install the package from Github, without anyother steps. You may be asked to update packages. This might not bestrictly necessary, so you can also try skipping this step.

Getting started

COINr needs a little reading and learning to understand properly. Butonce you have done that, it can be very powerful for developingcomposite indicators.

A good place to get started is COINr’s “Overview” vignette. Tryvignette("overview").

The most thorough documentation is available atCOINr’s website (developedusing pkgdown). This contains all package documentation in aneasy-to-navigate format. All documentation available here is alsoavailable by browsing COINr vignettes: seevignette(package = "COINr").

Tutorials

COINr was presented at the European Commission’strainingweek on composite indicators and scoreboards in September 2023,including a short one hour tutorial:

• See the presentation videohere(COINr presentation starts around 16:28 but if you are new to compositeindicators it is well worth following the full training)
• See the tutorial documenthereand the code in theGitHub repohere.

Recent updates

COINr has been recently updated to v1.0, skipping a few versionnumbers. This has brought in many new features, some discarded features,less dependencies and more robust underlying code. The syntax has alsobeen changed to make the package more consistent. Seevignette("v1") to learn about these changes if you wereusing COINr prior to v1.0.

COINr documentation was previously contained in anonline book. This isstill available, and although the principles of composite indicatorsthere are still all valid, the code refers strictly to COINr <v.1.0.

If you prefer to roll back to the old COINr, you can still install itas a separate package called “COINr6”. This is available on GitHub:

remotes::install_github("bluefoxr/COINr6")

Help and issues

For general help with COINr, the best place to look is the package’sdocumentation which is available either via the command line(vignette(package = "COINr")) or by checking individualfunction documentation (?function_name). All documentationis also conveniently available online atCOINr’s website.

If you find any problems with the package, including bugs orsuggestions, either open a GitHub issue here, or else contact me byemail.

Finally, contributions to the package are most welcome. This shouldbe done by cloning the repo, making your modifications, and then openinga pull request. You could also contact me in advance to discuss changesand extensions. Any changes (especially new functions) should beaccompanied by unit tests, and all existing tests should run withouterrors or warnings. To do this, run:

devtools::test()

Citing COINr

If you have found COINr helpful, we are grateful if you cite thepackage. COINr is citable by a paper in the Journal of Open SourceSoftware which you can findhere (withcitation information).

In R you can also generate the citation info usingcitation(package = "COINr"), or see the same info on theCOINr webpage.

Acknowledgements

COINr was initially developed under contract for the EuropeanCommission’s Joint Research Centre until 2021, and this enabled the bulkof the initial development, as well as many helpful discussions withcolleagues from theCompetenceCentre for Composite Indicators and Scoreboards, and this isgratefully acknowledged. Since then, the package has been continuouslymodified and improved while working on other projects and following userfeedback.