- A warning has been added to explicitly warn the userwhen LDA assumptions might not be met.
- Most of the internal functions are now exported, which means that
rdss
can now be used through R scripts. - A package vignette has been added to document the use of
rdss
through the command line. - An example data object (
poundbury
) has been included.
- Add a CITATION file.
- Updated references in documentation files.
Tiny changes in documentation and UI after acceptance of the associated article by the International Journal of Osteoarchaeology.
This is the first release of rdss
which is ready for public use. This version takes into account the suggestions made by anonymous reviewers of the International Journal of Osteoarchaeology, who must be acknowledged for their work.
- New option for the "Import data" step: individuals with all missing values can now be automatically removed.
- The whole data file can now be viewed in a pop-up just after the importation step.
- The UI widgets (filtering criteria) now reset to their default values when chosing a new target individual.
- New UI layout for tab "3. Check or customize the reference sample".
- The imputed reference dataset can now be downloaded through the user interface.
- More confidence levels (i.e., classification thresholds) are now allowed in tab "4. Perform sex estimation".
- Classification accuracy in LOOCV is now also given for females and males separately, in search for potential bias in sex estimation.
- Fixed typos in UI.
- Small code improvements and optimization.
- Additional tips and help have been added in the last tab of the user interface.
- A loading screen has been added for the computer-intensive method of missing data imputation.
- Some very simple unit tests have been added for the most elementary functions.
- Results of sensitivity analysis are now correctly displayed.
- Improved behavior of some UI widgets.
- Fixed issue when submitting filtered datasets to Robust LDA or Penalized regression.
- Fixed issue for displaying the results of robust LDA.
- An example file has been added.
- Robust linear discriminant analysis is now available for sex estimation.
- Small UI adjustments.
- Penalized logistic regression (ridge or LASSO) is now available for sex estimation.
- Improved auto-tuning for random forests.
- More parameters for random forests have been added in the UI.
- UI: large tables (if many variables are used for sex estimation) are now better displayed in the UI.
- UI: variable names can now be rotated on the
md.pattern()
plot.
First (experimental) release on GitLab.