- Additional logic in tests to verify that supplemental packages are available, skipping the test if said packages are not installed.
- Update tests to convert tibbles to data.frames prior to comparion as a work around for dplyr/2751.
- Previous versions sampled threat frequency (TEF) as a continuous distribution. Threat event counts are discrete (they either happen or don’t in a given simulated period), so this previous method was incorrect and inflated threat counts. The differences are small in the standard 10,000 evaluation range, but users should note the change.
- Part of the discretization step for TEF has a positive side effect of making the threat_event column in data objects an integer rather than a double- long. As this is still a numeric type, there should be no impact on users for this change apart from a slightly tidyer data output.
- Update risk_dashboard to not cache any chunks. This keeps
render from from trying to write to the package install directory.
- Updated tests to use four significant digits where long refrence values are passed, avoiding precission issues on machines without long doubles.
- Document optional dependency on pandoc and add test skip logic to avoid rmarkdown tests on systems where pandoc is not available.
- Add testing to travis matrix for Linux builds with and without pandoc.
- run_analysis script was missing namespace calls for
- All rmarkdown calls now specify an
intermediates_dir value of
tempdir(). This can be overwritten on the function call if needed.
- Report generation is now much quieter by default.
- New sample dataset:
mappings contains sample qualitative to quantitative parameters.
- risk_dashboard expects a mandatory output parameter to the desired rendered output.
pkgdown generated web documentation at https://evaluator.severski.net.
- Now accepts
format parameter to specify HTML, PDF, or Word
styles parameter allows user to supply custom CSS or Word reference document to customize styles and fonts
- RMarkdown 1.8.2 or greater (currently GitHub only) required to work around the temporary file deletion issue specified in https://github.com/rstudio/rmarkdown/issues/1184. Use
devtools::install_github("rstudio/rmarkdown", "b84f706") or greater.
- Expose OpenFAIR model selection in
- Provide default TEF/TC/DIFF/LM OpenFAIR model
create_templates function for populating starter/sample files, making starting a fresh analysis easier than ever!
- Experimental quick start script, run_analysis.R, supplied with
- All default directories normallized to ~/evaluator/[inputs|results]
- New OpenFAIR primitives:
- New composition functions:
- New difficulty composition functions:
- Improved help documentation on many functions
- Update of usage vignette
- Auto loads
extrafont database for better font detection
- Falls back to standard
sans family when none of the preferred options are available
- Drop use of
tcltk progress bar in favor of console-compatible
dplyr::progress_estimated. Also enables reduced package dependencies.
- Tests and code coverage reporting added
- Improve faulty
- Removed dependency on
generate_report defaults to creating a MS Word document as the output type
- Replaced dependency on
modeest with a slimmer
- Removed dependency on
- Default (overridable) locations of input and results directories now consistently set to “~/data” and “~/results” respectively
generate_report now takes an optional
focus_scenario_ids parameter to override the scenarios on which special emphasis (usually executive interest) is desired.
- Improve user experience for optional packages. User is now prompted to install optional dependencies (shiny, DT, flexdashboard, statip, rmarkdown, etc.) when running reporting functionality which requires them.
- Substantial improvements in the sample analysis flow detailed in the usage vignette. You can now actually run all the commands as-is and have them work, which was previously “challenging”.
summarize_all renamed to the more descriptive
summarize_to_disk to avoid dplyr conflict
- Add requirement for at least pander v0.6.1 for
- Substantial refactoring on vignette
- Added missing save steps
- Corrected package name for
- Fixed a few incorrect placeholders
- Properly committed compiled files to package for distribution and installation
- Update all tidyverse calls to account for deprecations and split out of
- Windows CI builds added via Appveyor
annotate_logticks over manual breaks on risk_dashboard
- Initial submission to CRAN