R package for creating statistical information packages to fulfil requirements of Executive Order 128

This UNOFFICIAL package can construct statistical information packages for submission to the Danish National Archives (Rigsarkivet). You require a package identifier and metadata files from the Rigsarkivet to do this.

This package is NOT ENDORSED BY THE RIGSARKIVET. However, it may be helpful.

Installation

The simplest way is

  library(devtools)
  install_github("BiomathematicsAndStatisticsScotland/rigsarkivet_r_infopkg/RigsArkivetRInfoPkg")

Relationship with ASTA

When exporting this tool performs various simple checks for data consistency and tries to warn the user of problems, it is ''not'' a complete package consistency checker. Any output from this tool ''must'' be checked using ASTA.

See the Rigsarkivet guidance pages for further information on using ASTA

Simple example

A simple invocation of the package will rebuild a version of the FD.18005 example data package from example data stored within the RigsArkivetRInfoPkg source directory:

    library(RigsArkivetRInfoPkg)
    pkg_output_dir=tempdir()

    pkg_description = list(
     archive_index=system.file("extdata", "FD_18005_archiveIndex.xml", package="RigsArkivetRInfoPkg"),
     context_doc_index=system.file("extdata", "FD_18005_contextDocumentationIndex.xml", package="RigsArkivetRInfoPkg"),
     pkg_id="FD.18005",
                   tables=list(
                       list(
                          name="R_testfil",
                          label_file=fd_18005_r_labels,
                          user_codes=list( gender='3',
                                           grade='3',
                                           education='6',
                                           hobby=c('8')
                                          ),                          
                          key_variable=c("child_id"),
                          description="Danish Longitudinal Research Study of Grandparents, Parents and Children - this is data 1",
                          table_dataset=fd_18005_r)))

    process_full_info_pkg(pkg_description,
                          output_dir = pkg_output_dir)   

Note that the emitted package will differ slightly from the original FD.18005 package. This is due to transforms applied to the base data in the package to make it more "R-like" - the categorical variables gender, grade, education, hobby and so on are transformed into R factors which results in the numeric values for these fields changing when re-emitted as a dataset.

Additionally, a simple command-line tool is available which emits a single RDS file as an information package. See ASTA_Convert.R in the Github repository.

For more complex invocations (e.g. where multiple data tables are involved) it is recommended to write a script which contains the description of your information package as shown in the example above.

The package vignette contains further examples of how to use this software to emit more complex information packages.

Customisation options

Relabelling the dataset

You can control relabelling operations by using the function relabel_dataset. This is called implicitly when the parameter label_file is supplied to process_full_info_pkg

Modifying processing of the dataset

By replacing the function parameters:

• preprocess_dataset
• sanitise_numeric
• sanitise_character

when calling process_full_info_pkg or process_as_asta_export_script, you can control the processing of the dataset, including variable cleanup and so on.

It is recommended to write a new function which encapsulates a call to default_preprocess_dataset as there's a lot of useful logic which you will probably want to reuse, in addition to any extra cleaning logic you may wish to add.

You can of course "pre-clean" your dataset yourself. It is entirely up to you.