Beyond sigmoids: How to obtain well-calibrated probabilities from binary classifiers with beta calibration
Open accessThe DOI matches its Crossref record; 14 of the 16 references it cites were found in a registry or an index, and 2 are not checked yet.
Abstract
For optimal decision making under variable class distributions and misclassification costs a classifier needs to produce well-calibrated estimates of the posterior probability. Isotonic calibration is a powerful non-parametric method that is however prone to overfitting on smaller datasets; hence a parametric method based on the logistic sigmoidal curve is commonly used. While logistic calibration is designed for normally distributed per-class scores, we demonstrate experimentally that many classifiers including Naive Bayes and Adaboost suffer from a particular distortion where these score distributions are heavily skewed. In such cases logistic calibration can easily yield probability estimates that are worse than the original scores. Moreover, the logistic curve family does not include the identity function, and hence logistic calibration can easily uncalibrate a perfectly calibrated classifier. In this paper we solve all these problems with a richer class of parametric calibration maps based on the beta distribution. We derive the method from first principles and show that fitting it is as easy as fitting a logistic curve. Extensive experiments show that beta calibration is superior to logistic calibration for a wide range of classifiers: Naive Bayes, Adaboost, random forest, logistic regression, support vector machine and multi-layer perceptron. If the original classifier is already calibrated, then beta calibration learns a function close to the identity. On this we build a statistical test to recognise if the model deviates from being well-calibrated.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“Beta calibration is equivalent to logistic calibration if 𝛽 1 = 𝛽 2 .8,9 Otherwise, it is more flexible than logistic calibration.”
From Calibrating machine learning approaches for probability estimation: A comprehensive comparison (2023) · cites this paper as [8]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“One approach that works well is to model the classwise scores with Beta distributions (Kull et al., 2017a(Kull et al., , 2017b)).”
From Classifier calibration: a survey on how to assess and improve predicted class probabilities (2023) · cites this paper as Kull et al., 2017b
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“Isotonic calibration 8,12 is one of the most commonly used calibration approaches by the machine learning community.64”
From Calibrating machine learning approaches for probability estimation: A comprehensive comparison (2023) · cites this paper as [8]
The cited source says what the claim attributes to it.
In the citing paper
“One approach that works well is to model the classwise scores with Beta distributions (Kull et al., 2017a(Kull et al., , 2017b)).”
From Classifier calibration: a survey on how to assess and improve predicted class probabilities (2023) · cites this paper as Kull et al., 2017a
The cited source says what the claim attributes to it.
In the citing paper
“Beta calibration is specifically designed for probabilistic classifiers and alleviates both drawbacks of Platt scaling (Kull et al., 2017a, b).”
From Classifier calibration: a survey on how to assess and improve predicted class probabilities (2023) · cites this paper as Kull et al., 2017a, b
2 more claims about this paper
Create a free account to see every claim other papers make about this one.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Scikit-learn: Machine Learning in Python
Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Müller, A., Nothman, J., Louppe, G., et al. (2012) · arXiv · doi:10.48550/arxiv.1201.0490 · Cited by 71,016 · cited 4×
“The second is the one implemented in Python’s Scikit-learn, based on Adaboost method SAMME (Zhu et al., 2009), referred to as Ada-S.”
MethodInfluentialThe WEKA data mining software: an update
Hall, M., Frank, E., Holmes, G., Pfahringer, B., Reutemann, P., & Witten, I. H. (2009) · ACM SIGKDD Explorations Newsletter · doi:10.1145/1656274.1656278 · Cited by 17,750 · cited 1×
“However, as the task is simply univariate logistic regression with feature s and label y , it can be solved using the standard logistic regression functionality in any machine learning toolkit, including WEKA (Hall et al., 2009) and Scikit-learn (Pedregosa et al., 2011).”
MethodFriedman, J., Hastie, T., & Tibshirani, R. (2000) · The Annals of Statistics · doi:10.1214/aos/1016218223 · Cited by 6,884
Predicting good probabilities with supervised learning
Niculescu-Mizil, A., & Caruana, R. (2005) · Proceedings of the 22nd international conference on Machine learning - ICML '05 · doi:10.1145/1102351.1102430 · Cited by 1,884 · cited 2×
“This was expected, since isotonic calibration is the only non-parametric method in this study and literature has shown that it needs more data than logistic calibration to fit a good calibration map (Niculescu-Mizil and Caruana, 2005).”
MethodHastie, T., Rosset, S., Zhu, J., & Zou, H. (2009) · Statistics and Its Interface · doi:10.4310/sii.2009.v2.n3.a8 · Cited by 1,875 · cited 1×
“The second is the one implemented in Python’s Scikit-learn, based on Adaboost method SAMME (Zhu et al., 2009), referred to as Ada-S.”
MethodTransforming classifier scores into accurate multiclass probability estimates
Zadrozny, B., & Elkan, C. (2002) · Proceedings of the eighth ACM SIGKDD international conference on Knowledge discovery and data mining · doi:10.1145/775047.775151 · Cited by 1,280 · cited 3×
“The scores of a sufficiently calibrated classifier can be simply thresholded at a threshold directly derived from the misclassification cost (Zadrozny and Elkan, 2002).”
BackgroundMethodRobust classification for imprecise environments
Provost, F., & Fawcett, T. (2001) · Machine Learning · doi:10.1023/a:1007601015854 · Cited by 1,250
Platt, J. C. (2000) · Advances in Large-Margin Classifiers · doi:10.7551/mitpress/1113.003.0008 · Cited by 1,156 · cited 3×
“Logistic calibration was proposed by (Platt, 2000) to reduce overfitting by introducing a strong inductive bias which considers only calibration maps of the following form: where γ, δ are real-valued parameters with γ ≥ 0 to ensure that the calibration map is monotonically non-decreasing.”
MethodStatistical Comparisons of Classifiers over Multiple Data Sets
DemšarJanez (2006) · Journal of Machine Learning Research · doi:10.5555/1248547.1248548 · Cited by 1,074
Calibration of Probabilities: The State of the Art
Lichtenstein, S., Fischhoff, B., & Phillips, L. D. (1977) · Decision Making and Change in Human Affairs · doi:10.1007/978-94-010-1276-8_19 · Cited by 483
Cited by
Items 101–138 of 138
Using risk prediction models to inform personalized, cost-effective treatment recommendationsPreprint
Neves, M. R., Franke, M. F., Mitnick, C. D., Ciobanu, N., Crudu, V., Furin, J. J., Cohen, T., & Yaesoubi, R. (2025) · medRxiv · doi:10.1101/2025.08.07.25333118
“We used the Platt scaling [20] and the beta calibration [21, 22] methods to calibrate the model predictions.”
Assessing the capability of supervised classifiers to recover the true underlying probabilities
Dieye, N. A., Russolillo, G., & Niang, N. (2025) · International Conference on Advanced Machine Learning and Data Science (AMLDS) · doi:10.1109/amlds63918.2025.11159360
“We also use decompositions of BS and LL into components that isolate calibration quality (see Table I) [22], [23].”
Research on a Calibration Model for ML-Based Obfuscated Malware Detection
Jamgharyan, T. V. (2025) · Computer Science and Information Technologies (CSIT) · doi:10.51408/csit2025_41
“Unlike parametric approaches, isotonic regression does not impose assumptions on the shape of the calibration function [13], making it particularly suitable for tasks characterized by high uncertainty and noisy feature spaces.”
Uncertainty Calibration for Deep Neural Network
Liu, J. (2025) · ANU Open Research (Australian National University) · doi:10.25911/nxfc-sg97
Measuring the Error: Validation of Learning Models
Biscarini, F., & Nazzicari, N. (2026) · Springer eBooks · doi:10.1007/978-3-031-96852-5_3
Hybrid Generative–Predictive Models for Customer Risk Profiling in Insurance
Rajendran, S. (2025) · International Journal of Science and Research Archive · doi:10.30574/ijsra.2025.17.3.3211
Basha, M. S., Mahalakshmi, R., Lakshmi, P. M., Usha, S., Sheriff, R. A., & Lakshmipratha, V. (2025) · IEEE International Conference on Emerging Trends in Computing and Communication (ETCOM) · doi:10.1109/etcom66606.2025.11437030
“It predicts the probability of an instance belonging to a specific class by fitting a sigmoid function to the data [18].”
Cendre, R. (2021) · theses.fr (ABES) · doi:10.70675/92fd40b8z03c6z4261z88dez1d4b85e069a4 · Cited by 1
Rajaonison, A. (2021) · theses.fr (ABES) · doi:10.70675/0d3f077dzc92az4835z99b2za73fb34f82a7
Apprentissage actif multi-labels pour des architectures transformers
Arens, M. (2024) · theses.fr (ABES) · doi:10.70675/47cfa21dz027fz444ez86d6z2e820b15c5ef
Characterization of a Reliability Domain for Image Classifiers
Coz, A. L. (2024) · theses.fr (ABES) · doi:10.70675/b6f39077z9dc3z4edcz9e4cz073207f40c00
Knowledge Distillation for Wearable-Based Disease Prediction and Vital-Sign Classification
Song, J., Zhang, J., & Li, R. (2026) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp55912.2026.11460751
Ulyanin, M. Y. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.6625797
Identifying Courses for Targeted Review Using GAP Analysis and Machine Learning
Joseph, K., Calvert, W., Keys, O., & McCrocklin, S. (2026) · Education Sciences · doi:10.3390/educsci16050806 · Cited by 1
“Post-hoc global beta calibration (Kull et al., 2017) served as a robustness check, marginally improving probability calibration while resulting in minimal changes to unadjusted GAP course rankings.”
Output Correction of Recurrence-Aware Long-Term Cognitive Network Classifiers
Nápoles, G., Grau, I., & Salgueiro, Y. (2026) · Big Data and Cognitive Computing · doi:10.3390/bdcc10060178
“While Platt correction uses a simple affine transformation followed by a sigmoid, Beta correction incorporates a logarithmic curvature term b k log z 2 k + δ that allows the transformation to adapt more precisely to the shape of the raw score distribution [51].”
Research Model for Improving the Detection Accuracy of Obfuscated Malware
Jamgharyan, T. V. (2026) · Programming and Computer Software · doi:10.1134/s036176882670009x
“Unlike parametric approaches, isotonic regression does not impose assumptions on the shape of the calibration function [13], making it par-PROGRAMMING ticularly suitable for problems characterized by high uncertainty and noisy feature spaces.”
Calibration and Clinical Utility for Trustworthy Handwriting-Based Alzheimer's Disease Detection
Kim, Y.-I. (2026) · Journal of Software Forensics · doi:10.29056/jsf.2026.06.12
Quality-Aware Robust Scheduling for LPBF In718 Using Am-Bench 2022 Measurements
Sundana, S., & Ali, P. R. (2026) · Eksploatacja i Niezawodnosc - Maintenance and Reliability · doi:10.17531/ein/220968
SOULEIMAN, I. A., & Souleiman, H. A. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7082155
probcal: Calibration of Binary and Multiclass Probabilities
Marinho, P. R. D. (2026) · doi:10.32614/cran.package.probcal
Calibrating Machine Learning Approaches for Probability Estimation Without Calibration Data
di Carluccio, E., Koliopanos, G., Ojeda, F. M., Weimar, C., & Ziegler, A. (2026) · Statistics in Medicine · doi:10.1002/sim.70751
Tang, Z., Sun, C., Ren, X., Zhang, X., Zhu, J., & Li, H. (2026) · Sensors · doi:10.3390/s26144487
“The ROC curve, PR curve, Brier Score, expected calibration error, and calibration curve have been widely used to evaluate model discrimination ability and probability calibration quality [47–51].”
Ratchaneekorn, K., & Narkbunnum, W. (2026) · Information · doi:10.3390/info17070702
“Where calibration is poor, post hoc recalibration (Platt scaling and isotonic regression) is applied within the CV protocol; beta calibration provides parametric generalization [39,40].”
Severity-Weighted Calibration Error for Reliability Assessment Under Outcome-Severity Imbalance
Chandra, H. A., & Narendra, K. (2026) · IEEE Access · doi:10.1109/access.2026.3717951
“Kull et al. proposed beta calibration as a more flexible parametric alternative [7].”
Breast cancer risk prediction from longitudinal mammography reports in a real-world health systemPreprint
Monteiro, H. S., da Cruz, J. A. S., Mattos, C. L. C., Chaves, I., Machado, J. C., Makse, H. A., Parra, L. C., & Andrade, J. S. (2026) · Research Square · doi:10.21203/rs.3.rs-10168412/v1
Lee, Y., Hong, S., Jeong, J., Kim, D. R., Lee, C. H., & Surowiec, R. K. (2026) · Advanced Sensor Research · doi:10.1002/adsr.70188
“Without proper calibration, classifiers may output overconfident or poorly calibrated probability estimates, potentially triggering false alarms or missed detections in wearable applications [ 39, 40 ].”
Calibrated and Uncertainty-Aware Heart Disease Prediction Using Ensemble Learning
Dudeja, P., Mishra, D., Mathur, R., Gupta, S., Kumar, D., & Hemanth, D. J. (2026) · International Conference on Intelligent Engineering and Next-Gen Healthcare Systems (IEHNS) · doi:10.1109/iehns68708.2026.11606791
“…Forest, and Neural Network give very different probability predictions for the same patient, Instead, it represents a signal of epistemic uncertainty, which suggests that the patient is likely to be close to the boundary or that the feature sets are under-represented in the training data [3].”
SoftMCC: An MCC-Brier Calibration Bridge for Threshold-Free Model Selection under Class ImbalancePreprint
Canay, Ö. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2608.08984
“…2007), and calibration methods such as Platt scaling, isotonic regression, beta calibration, and modern neural-network temperature scaling are used to recalibrate probability estimates (Guo et al., 2017; Kull, Silva Filho, & Flach, 2017; Niculescu-Mizil & Caruana, 2005; Zadrozny & Elkan, 2002).”
翁, 长. (2026) · Journal of Sensor Technology and Application · doi:10.12677/jsta.2026.145083
Dickerson, J., Shinitzky, H., Peer, M., Ben‐Shachar, M. S., & Gilead, M. (2026) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2615627123
Bayati, M., & Eskandari, F. (2026) · Network Modeling Analysis in Health Informatics and Bioinformatics · doi:10.1007/s13721-026-00905-1
“Previous studies have similarly noted that boosted classifiers may require explicit calibration assessment despite strong classification performance (Niculescu-Mizil and Caruana 2005; Guo et al. 2017; Kull et al. 2017).”
Adapting Skew-Probabilistic Neural Networks for Credit-Risk Probability Estimation
Bimoldan, M., & Kasenov, S. E. (2026) · Machine Learning and Knowledge Extraction · doi:10.3390/make8110352
Calibration Results for Established Post-hoc Calibration Methods
Huang, W. (2025) · Zenodo (CERN European Organization for Nuclear Research) · doi:10.5281/zenodo.15868156
Ayoub, R., Haider, U., & Creutzburg, R. M. (2026) · International Conference on Computer Science and Engineering (UBMK) · doi:10.1109/ubmk71847.2026.11745391
“This design follows the calibration-under-imbalance literature [23]–[25] while extending it to the IDS governance context.”
Gao, S., Jia, T., Xu, Q., Tian, X., & Gang, S. (2026) · Water · doi:10.3390/w18212723
Post-Hoc Calibration Methods for Machine-Learning Forecasts of Football Outcomes
Roque, P. M. G. (2026) · Portuguese National Funding Agency for Science, Research and Technology (RCAAP Project by FCT)
Bekmurat, O., Sansyzbay, K., Chinibayev, Y., Bakhtiyarova, Y., Тасболатова, Л., Kismanova, A., & Nurmukhanova, A. (2026) · IoT · doi:10.3390/iot7040099
Prior-Invariant Max-Hellinger Forests For Skew-Insensitive Classification of Rare Events
Padmavathi, J., & Burugari, V. K. (2026) · Adolescencia e Saude · doi:10.67440/ahj.vi.3125
Metadata
- Authors
- Kull, M., Silva Filho, T. M., & Flach, P.
- Affiliations
- University of Tartu (EE)
- University of Bristol (GB)
- Venue
- Electronic Journal of Statistics
- Publisher
- Institute of Mathematical Statistics
- Published
- 2017-01-01
- Volume
- 11
- Issue
- 2
- Pages
- 5052-5080
- ISSN
- 1935-7524 (ISSN-L)
- OpenAlex
- W2773235741
- Cited by
- 111
- References
- 15
- Citations by year
- 2026: 22 · 2025: 16 · 2024: 11 · 2023: 16 · 2022: 10 · 2021: 17 · 2020: 11 · 2019: 7 · 2018: 1
- FWCI
- 3.10
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- cc-by
- Locations
- Electronic Journal of Statistics (publishedVersion, cc-by)
- Explore Bristol Research (submittedVersion, cc-by)
- Project Euclid (Cornell University) (submittedVersion)
- Bristol Research (University of Bristol) (submittedVersion, cc-by)
- Funders
- Eesti Teadusagentuur (PUT1458)
- Conselho Nacional de Desenvolvimento Científico e Tecnológico
- Engineering and Physical Sciences Research Council (EP/K031910/1, EP/K018728/1)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Imbalanced Data Classification Techniques
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning and Data Classification
- Physical Sciences › Computer Science › Artificial Intelligence › Adversarial Robustness in Machine Learning
- Keywords
- probability calibration, binary classifiers, posterior probability estimation, isotonic calibration, naive Bayes, AdaBoost, random forest, support vector machine, multilayer perceptron, calibration tests
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.