Package: probcal 0.1.0

probcal: Calibration of Binary and Multiclass Probabilities

Provides S3 calibrators, metrics, and diagnostics for binary and multiclass probability calibration in R. Binary methods include Platt scaling, temperature scaling, beta calibration, histogram binning, and isotonic regression. Multiclass methods include temperature scaling, vector scaling, Dirichlet calibration, and a one-vs-rest wrapper for the binary calibrators. Methods follow Platt (1999), Zadrozny and Elkan (2002) <doi:10.1145/775047.775151>, Guo et al. (2017), Kull et al. (2017) <doi:10.1214/17-EJS1338SI>, and Kull et al. (2019).

Authors:Pedro Rafael Diniz Marinho [aut, cre]

probcal_0.1.0.tar.gz
probcal_0.1.0.zip(r-4.7)probcal_0.1.0.zip(r-4.6)probcal_0.1.0.zip(r-4.5)
probcal_0.1.0.tgz(r-4.6-any)probcal_0.1.0.tgz(r-4.5-any)
probcal_0.1.0.tar.gz(r-4.7-any)probcal_0.1.0.tar.gz(r-4.6-any)
probcal_0.1.0.tgz(r-4.6-emscripten)
manual.pdf |manual.html✨
DESCRIPTION |NEWS
card.svg |card.png
probcal/json (API)

# Install 'probcal' in R:
install.packages('probcal', repos = c('https://prdm0.r-universe.dev', 'https://cloud.r-project.org'))

Bug tracker:https://github.com/prdm0/probcal/issues

Pkgdown/docs site:https://prdm0.github.io

On CRAN:

Conda:

machine-learningrlanstatistical-learningstatistics

3.51 score 13 scripts 16 exports 17 dependencies

Last updated from:fd04d574ab. Checks:9 OK. Indexed: yes.

TargetResultTimeFilesSyslog
linux-devel-x86_64OK146
source / vignettesOK200
linux-release-x86_64OK140
macos-release-arm64OK111
macos-oldrel-arm64OK133
windows-develOK100
windows-releaseOK133
windows-oldrelOK88
wasm-releaseOK130

Exports:acecal_betacal_cvcal_dirichletcal_histogramcal_isotoniccal_ovrcal_plattcal_temperaturecal_vector_scalingeceinv_logitlogitmcemmcereliability_diagram

Dependencies:clicpp11farverggplot2gluegtableisobandlabelinglifecycleR6RColorBrewerrlangS7scalesvctrsviridisLitewithr

Applied Calibration Workflow
Goal | Prepare the data | Fit a classifier | Fit calibrators | Compare calibration metrics | Plot the calibrated probabilities
Rendered fromapplied-workflow.Rmdusingknitr::rmarkdown

Last update: 2026-06-23
Started: 2026-06-23

Calibrating Binary Probabilities
Why calibration matters | A three-split workflow | Fit a calibrator | Compare methods | Reliability diagram | Cross-validated calibration | Optional reference validation | Current scope
Rendered fromprobcal.Rmdusingknitr::rmarkdown

Last update: 2026-06-23
Started: 2026-06-23

Choosing a Calibrator
The main decision | Match the input scale | Compare methods on held-out data | Practical guidance
Rendered fromchoosing-a-calibrator.Rmdusingknitr::rmarkdown

Last update: 2026-06-23
Started: 2026-06-23

Multiclass Calibration
From two classes to several | Simulating an overconfident classifier | Measuring multiclass calibration | Temperature scaling on logits | Dirichlet calibration on probabilities | One-vs-rest calibration | Comparing the calibrators | Reliability diagram | Out-of-fold calibration | Scope
Rendered frommulticlass.Rmdusingknitr::rmarkdown

Last update: 2026-06-23
Started: 2026-06-23

Numerical Validation
Purpose | Optional checks | Why not compare every method | Running the optional tests
Rendered fromnumerical-validation.Rmdusingknitr::rmarkdown

Last update: 2026-06-23
Started: 2026-06-23