0.9.3

This is largely a patch release dealing with processing issues introduced by the recent release of tidyr 1.2.0. In addition support for data from Estonia and Vietnam has been added. See the full release notes for details. Thanks to @RichardMN, @biocyberman, @Bisaloo, @sbfnk, and @kathsherratt for contributing to this release.

New data sets

• Support for level 1 region data in Estonia (thanks to @RichardMN). See ?Estonia for details.
• Support for level 1 region data in Vietnam (thanks to @biocyberman). See ?Vietnam for details. Note that this class is only available if installing the vietnam branch from GitHub due to API instability issues.

Other changes

• Change the data source for Switzerland to draw data from the Swiss Federal Office of Public Health (FOPH)
• Updated the package logo to include the newly supported data sets.
• Reduced the number of package dependencies (@Bisaloo and @RichardMN)
• Added tests for JSON download code (@RichardMN).

Bug fixes

• Fixed a bug in fill_empty_dates_with_na() caused by changes made in version 1.2.0 of tidyr.
• Fixed a bug in the data sourced from Germany so that instead of treating it as a line list of individuals it is treated as a relatively finely resolved count data which needs to be summed up (by @sbfnk).
• Fixed a bug in the Vietnam class due to stringr (#448 by @RichardMN).
• Fixed a bug with the Netherlands class were the lack of Hospitalisation data in the source was causing the class to fail (#446 by @RichardMN).
• Fixed an issue with the Colombia data and reduced dependencies by making RSocrata be a suggested package (#433 by @RichardMN).

Depreciations

• get_interventions_data() has been depreciated and removed.
• get_info_covidregionaldata() has been depreciated and removed.
• The country argument to get_national_data() has been depreciated and removed. It's functionality can be replicated using the countries argument.
• get_linelist() has been updated to fail on use vs warn on use and will be removed from future versions of the package.

What's Changed

• Fixing paper formatting. by @arfon in #395
• fixed Richards last name by @joseph-palmer in #396
• added link to discord chat in readme badge by @joseph-palmer in #401
• Fixes citation by rebuilding readme by @seabbs in #402
• Make bot less bossy by @seabbs in #407
• Add Estonia data by @RichardMN in #417
• Add subnational data for Vietnam by @biocyberman in #413
• Update vietnam url and docs by @RichardMN in #428
• Update stable API URL of Vietnam by @ilumitr in #427
• fix Germany case/death data by @sbfnk in #436
• Remove action re-building JOSS paper by @Bisaloo in #438
• Fix giant logo by specifying width instead of height by @Bisaloo in #442
• Switch to bootstrap 5 for pkgdown by @seabbs in #443
• New data source for Switzerland - fix for #412 by @RichardMN in #429
• Update pr-checklist.yaml to include news nudge by @seabbs in #439
• Reduce the number of Imports by @Bisaloo in #437
• Fix website URL by @Bisaloo in #444
• Hugo: ctb -> aut by @seabbs in #416
• Don't rely on exact column ordering when grouped data is involved by @DavisVaughan in #445
• Fix lintr warnings by @Bisaloo in #450
• Bug 454 countrycode by @seabbs in #455
• Bug #449 tidyr complete by @seabbs in #453
• Refactor Vietnam cleaning code and change string conversion function - fixes #447 by @RichardMN in #448
• Depreciations by @seabbs in #456
• Add tests for json_reader function and sample json data by @Bisaloo in #451
• Adjust for Netherlands - remove Hospitalisation data - partial fix for #430 by @RichardMN in #446
• Alternate fix colombia data 430 which makes RSocrata suggested by @RichardMN in #433
• Bug: #460 Colombia failing tests by @seabbs in #461

New Contributors

• @arfon made their first contribution in #395
• @biocyberman made their first contribution in #413
• @ilumitr made their first contribution in #427
• @DavisVaughan made their first contribution in #445

Full Changelog: 0.9.2...0.9.3

JOSS release

This release adds support for the Covid19 Data Hub which includes Google and Apple mobility data amongst a large range of other data sets, data from the European Commission's Joint Research Centre which is at both the regional and national level, and individual sources for regional data from several countries. Package updates have been made in line with a software review at the Journal of Open Source Software. Finally, this release exposes more of the testing infrastructure to users and adds a package hexsticker.

Thanks to @joseph-palmer, @RichardMN, and @kathsherratt for contributions towards this release.

New features

• Support added for data sets from Covid19 Data Hub. This source aggregates a range of data at a national and subnational level and provides keys to link to mobility data provided by Apple and Google (by @joseph-palmer).
• Support added for data from the European Commission's Joint Research Centre (JRC). The source aggregates incidence data at the country and regional level for 34 UCPM Participating States plus Switzerland (by @joseph-palmer).
• Support added for data from the Netherlands provided by RVIM (English: National Institute for Public Health and the Environment). This source provides case, deaths and hospital admission data at the province and municipal levels (by @joseph-palmer).
• Support added for data from Switzerland and Liechtenstein collated by Canton Zurich (@openZH). This source provides case, deaths and hospital admission data at the canton level (by @RichardMN).
• Made package changes recomended in the JOSS review, including additional statements of need to the README, updates to the manuscript (paper.md) and fixes a bug of multiple sources for some countries. We are very grateful for the detailed feedback given by the JOSS reviewers and their help in improving this package.

Changes to implemented data sources

• Increased the robustness of fetching UK NHS admissions by region. Rather than testing a single date for data we now look over the last 7 days and pick the most recent available data set (by @kathsherratt).

Other changes

• Testing of classes updated to allow for at least one of common_data_urls or level_data_urls to be present. The previous default which forced the presence of common_data_urls meant that several classes had to define an empty field (by @joseph-palmer).
• Tests on data sets are now included as a method in DataClass. test_regional-datasets now calls the test function for all classes at each level. Data set specific tests (such as for NHS regions in the UK) are included as a specific_tests function within the country class, which is called by the parent (DataClass) test after performing standard checks. This allows all the code about a country to be defined in its own class. In addition, users can run tests interactively by calling the test method (e.g. $test()) (by @joseph-palmer)
• A function to create a template class and automatically add a github workflow file has been added. This makes adding a new data source for a country even easier as now you can call the function make_new_data_source() with the country / source name to add and it will set up the basic structure for you. There is also now a github check to make sure all new sources have a workflow set up (by @joseph-palmer).
• Adds source_ fields to all data sets to help users properly attribute their data sources (by @RichardMN).

Bug fixes

• An issue where the Lithuania() data set would ignore optional class specific arguments has been fixed (by @RichardMN).
• An issue where the JHU() source had multiple region codes for each country has been fixed, giving just one region code per country (by @joseph-palmer).

0.9.1: JHU and Google support

This release adds support for data sets from John Hopkins University and the Google open data project. Both of these sources aggregate a range of data at national and subnational levels. It also contains a range of small fixes and improvements to documentation. Finally, this release adds optional data processing which will be extended in future releases (contributions warmly welcomed).

Thanks to @joseph-palmer, @RichardMN, and @kathsherratt for contributions towards this release.

New features

• Support for data provided by John Hopkins University (by @joseph-palmer).
• Support for data provided by Google COVID-19 open data project (by @joseph-palmer).
• Added a available_regions method for all classes that shows level 1 regions with data available for the region of interest. This is of particular use when combined with the JHU or Google datasets where processing a large number of regions that are not required can take some time.
• Adds support for JHU or Google data to get_national_data(). This may also now be used to access lower level data from these sources but it may be better to instead use the classes directly or via initialise_dataclass().

Other changes

• The optional downloading of NHS region data in the UK() has been improved to include both the dynamic data previously supported and the archive document now produced (by @kathsherratt).
• The examples for the UK() class have been expanded to better showcase the package functionality.
• The documentation and examples for get_regional_data(), get_national_data(), and get_available_datasets() has been expanded with a focus on increasing the visibility of the underlying package structure.
• The documentation and examples for initialise_dataclass(), DataClass(), and CountryDataClass() has been expanded and improved.
• Improvements to the linking of documentation for related functions and classes.
• Improvements to the documentation for contributors (by @RichardMN).
• Improvements to the pkgdown documentation to organise packages into appropriate subcategories.

0.9.0 CRAN release

In this release covidregionaldata has been substantially retooled to be more robust, and to handle data in a more transparent way. Adding new data sets and functionality has also been made more streamlined. As this update is a substantial package refactor some breaking changes may be been inadvertently introduced. If requiring the old behaviour please install [email protected] from GitHub.

Thanks to @joseph-palmer, @RichardMN, and @kathsherratt for major contributions to this release. Thanks to @RichardMN for volunteering his time.

New features

• Track data processing from raw to clean using the step = TRUE argument in get_regional_data().
• Filter datasets for regions and countries of interest.
• Access the underlying methods for data sets and all steps in the data processing pipeline.

Documentation

• All vignettes have been updated for readability.
• A quickstart has been added to the package README.

Technical improvements

• get_regional_data() and get_national_data() now use R6 method dispatch. This is an internal change and so should have minimal user impact for users of the get_ functions. However, all datasets are now available to be used as R6 methods (see get_available_datasets) which may allow for more modular use cases. These classes can also be initialised using initialise_dataclass() which is used internally by both get_regional_data() and get_national_data().
• Unit testing has been separated from data downloading which is now tested individually by data set. This allows for contributors to more easily assess the impact of their additions and also allows us to publish data status updates for each data sets (see the README: https://github.com/epiforecasts/covidregionaldata#readme).

Deprecated functions

• get_available_datasets() replaces get_info_covidregionaldata() to view available data. get_info_covidregionaldata() is deprecated.
• get_interventions_data() is deprecated. These data no longer update as of December 2020. Check for alternatives at https://supertracker.spi.ox.ac.uk/policy-trackers/
• get_linelist is deprecated. Linelist stopped updating June 2020. Up to date linelist data are now behind a login: access at https://global.health/. We are working on a solution for accessing with covidregionaldata.

Data changes since 0.8.3

• Colombia now has capitalized region names.
• Germany level 2 region codes have been removed (previously was all NAs).
• India uses NA for unknown region codes, a change from IN-UN previously.
• Italy column region is now regioni.
• Mexico codes 'inegi_code' has been changed to 'inegi'.
• UK Level 1 'ons_region_code' is now 'region_code'.
• UK level 2 "ltla_code" is now "local_authority_code".
• get_available_datasets() now return an origin column rather than a country column and a type column rather than a get_data_function to better reflect the types of data supported.

Archive release of pre-refactor package

Merge pull request #248 from epiforecasts/keep_na

keep NAs when putting together data (in the UK)

Pre-CRAN release

Testing release.

Data source updates

This release contains fixes to keep data sources current and adapt to upstream changes but adds no new features.

Stable release (pre-CRAN)

Merge pull request #82 from epiforecasts/brazil

Brazil - fix broken link

New CRAN release

• Added whitespace trimming to all regional data functions.
• Fixed region codes for Colombia.
• Fixed region name cleaning for afghanistan.
• Updated UK data source and expanded available variables based on the newly implemented API.
• Enabled regional localisation to be optional.
• Minor quality of life changes.

Initial CRAN release

Merge pull request #96 from epiforecasts/uk-regional-patch

Update to get_uk_regional_cases