- removed Unicode characters from plot names (as requested by CRAN)
- fixed two instances of class(x) == "sth" instead of inherits
- more stable AUC computation #157
- updated documentation
- bugfix:
score_auc() - add spellcheck
- fix spelling
- markdown added to suggests
- new default color palette
DALEX::colors_discrete_drwhy - depend on
R v3.5to comply withDALEX - Allow for some deprecated names in
plot()function, to make it compatible with the paper auditor: an R Package for Model-Agnostic Visual Validation and Diagnostics in The R Journal, - add
arrow_sizetoplot_pcaand increase the default width value to2 - add
show_rugstoplot_residual_densitywhich allows to hide the rugs layer - add
score_auprcwhich calculates the AUPRC measure - add
plot_prcwhich plots the Precision-Recall Curve - use
stringsAsFactors=TRUEwhen creatingdata.frameinmodel_*()functions - add
coord_fixedtoplot_coord_fixed - add
predict_funcionandresidual_functiontoauditfunction
- Change
scoreparameter totypeinscore()function - Fix a bug in calculating Cook's distances
- Fix a bug in plotting Lift Charts
- Add
dataandcutoffparameter to severalscore_xfunctions. - Add
score_gini,score_one_minus_gini, andscore_r2functions
- Szymon Maksymiuk is now contributor
- New score functions:
score_acc,score_f1,score_precision,score_recall,score_specificity,score_one_minus_acc,score_one_minus_auc,score_one_minus_f1,score_one_minus_precision,score_one_minus_specificity,score_one_minus_recall
- Tomasz Mikołajczyk, Hubert Baniecki, and Michał Burdukiewicz are now contributors :)
- New names of functions #95, the old ones are now depreciated.
- All plots gained new drWhy theme.
modelEvaluation()return one consistent data frame.- Unnecessary Imports and Suggests removed from DESCRIPTION.
- Interactive D3 plots were added.
- Repository with the
auditorpackage was moved to new GitHub organization (ModelOriented).
- Automated checks for residuals are added. See the
check_residuals()function.
- Repository with the
auditorpackage was moved to new GitHub organization (MI^2 DataLab).
- New five vignettes with examples.
- Second pipeline with new computational functions:
modelResiduals(),modelEvaluation(),modelFit(),modelPerformance(),observationInfluence(). - New plot:
plotResidualBoxplot()
- In the source code, function
isFALSE()is replaced by== FALSEto make package compatible with R versions older than 3.5.