This vignette outlines the design decisions that have been taken
during the development of the {epiparameterDB} R package,
and provides some of the reasoning, and possible pros and cons of each
decision.
This document is primarily intended to be read by those interested in understanding the code within the package and for potential package contributors.
{epiparameterDB} provides a database of epidemiological
parameters available in R. It is the data component of the
{epiparameter} R package, and is in a separate R package to
appropriately license the data (stored in this package) and the code for
working with epidemiological parameters (stored in
{epiparameter}).
This R package only exports data, it does not export any R functions. It contains a few internal functions that are used for data processing and vignettes.
The epidemiological parameter database is exported by the package as
epiparameterDB::epiparameterDB. The
epiparameterDB data object is a list for flexible
hierarchical parameter data.
The raw epidemiological parameter data is stored within this package
as a .json file to enable structured and nested data
objects and the file is read into R using
jsonlite::read_json() in the .read_db()
internal function.
The epiparameterDB data object is then used within the
{epiparameter} package and converted to a list of
<epiparameter> objects, with a
<multi_epiparameter> class attribute to enable custom
methods for multiple <epiparameter> objects. This
conversion and class specification all takes place in
{epiparameter} to prevent duplicating code across packages
and to keep {epiparameterDB} to remain a minimal and
data-focused as possible.
The package exports the raw JSON list not a
<multi_epiparameter> formatted object due to the
dependencies (e.g. {distributional} and
{checkmate}) and {epiparameter} functions
(e.g. create_prob_distribution()).
.cite_author() is a simpler version of the internal
.citet() function in {epiparameter}. In
{epiparameterDB} the parameter database is validated by the
json-validate.yaml workflow so the format is consistent and
the individual elements for the short citation in the
database.Rmd vignette can be extracted and only the author
formatting needs to be packaged into a function.
the <epiparameterDB> class attribute attached
to the epidemiological parameter list is to ensure the
parameters.json file has been read using
.read_db() and suggests the file has not be read or
invalidated using other functions. It has no role in for object-oriented
programming (e.g. class methods).
The {epiparameterDB} package has no imported (i.e. hard)
dependencies. It is a data package and does not require any imported
functionality.
There are a handful of suggested (i.e. soft) dependencies. These enable the data to be read from file and to render the vignettes.
{jsonlite}:
used by .read_db() to load the parameters.json
parameter library into memory.{DT}: used
by the database.Rmd to render a DataTables on the
vignette.There are some additional suggested dependencies for vignette
rendering and testing the package: {knitr},
{rmarkdown}, {spelling},
{testthat}.
The data exported in this package was original included in the
{epiparameter} R package. Due to CRAN regulations, it was
not possible to dual license {epiparameter} with MIT and
CC0 for the code and data respectively. Therefore,
{epiparameterDB} is the data storage and distribution
component of the {epiparameter} project.