Pith. sign in

REVIEW 4 major objections 5 minor 40 references

Clustered Calibration: Representation-Aware Probability Calibration via Learned Subpopulations

T0 review · 4 major / 5 minor · reviewed 2026-08-04 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper argues that cluster-specific calibration beats a single global calibration curve and that its cluster-based error metric gives a model-selection rule that optimizes both calibration and discrimination.

desk verdict The empirical idea is worth a look, but the headline guarantee—lower CECE implies higher AUC—is false, and the paper should not be accepted on that claim. read the letter →

arxiv 2510.19328 v2 pith:PT7SUAO7 submitted 2025-10-22 cs.LG

Tomer Lavi , Bracha Shapira , Nadav Rappoport This is my paper
classification cs.LG
keywords probabilitycalibrationexpectederrorclusteredmodelselectionROC-AUClearnedrepresentationssubpopulationsuncertaintyquantification
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper argues that a classifier's predicted probabilities can be made more reliable by first clustering the data in a learned representation of the inputs, so that each cluster approximates a subpopulation the model treats differently, and then fitting a separate calibration curve to each cluster instead of one global curve. It introduces CECE, a calibration metric whose bins are those learned clusters, and claims that CECE is the only common metric that lets model selection optimize both calibration and discrimination: a lower CECE provably implies a higher ROC-AUC, a standard measure of ranking ability. If this holds, calibration and ranking stop being competing objectives: improving calibration on the right subpopulations also improves the model's ability to separate positive from negative cases.

What carries the argument

The central object is CECE, an expected-calibration-error variant computed over cluster bins defined by a learned representation rather than over confidence intervals. Because the bin assignment depends only on the sample's embedded representation, not on the calibrated output, the partition is invariant to the calibrator; that invariance is what lets a CECE comparison between calibrators translate into a ranking of their ROC-AUC. The companion CCL procedure is the ensemble of cluster-specific calibrators that this metric evaluates, and the paper shows that such clusters can reveal meaningful subpopulations that are not defined in advance.

What would settle it

Take one dataset and one cluster partition; fit both a global calibrator and cluster-specific calibrators of the same family, minimizing ECE directly on the calibration split. If any cluster's global-calibrator ECE is below the cluster-specific calibrator's ECE, Lemma 2.1 is contradicted; a synthetic mixture with two well-separated Gaussian components and different label rates should be the easiest place to search.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that shifting the binning of calibration error from predicted-confidence intervals to learned subpopulations changes what the metric can certify. CCL trains an ensemble of cluster-specific calibrators, and CECE evaluates them on the same clusters. The authors prove that, under an optimal-fit assumption for the calibrators, CECE bounds the error of any global calibrator from above and that a strict CECE improvement guarantees a strict ROC-AUC improvement, so the cluster-based metric can be used for model selection where standard ECE cannot, because standard ECE's bins shift when the calibrator changes.

Load-bearing premise

The chain of guarantees rests on the assumption that the fitted calibrator for each cluster is the one that truly minimizes calibration error on that cluster; the implementations reported in the paper fit their parameters by log-loss, not by calibration error, so a global calibrator can in principle beat a cluster-specific one on that cluster's data.

Editorial extensions

If this is right

  • If CECE is a valid model-selection criterion, practitioners can tune calibration methods without a separate discrimination check, since lower CECE would certify better ranking as well.
  • Cluster-specific calibration can identify unknown subpopulations, such as rare clinical subgroups, and correct their calibration errors, adding interpretability to the calibration step.
  • CCL is a template method that can be applied on top of any base classifier, any learned representation, any clustering algorithm, and any parametric calibrator, giving it broad applicability.
  • In rejection and abstention settings, choosing the CCL variant with the best CECE should produce the lowest error on accepted samples, supporting safer deferral to human experts.
  • The paper's argument that fixed-bin ECE can mis-rank soft, region-aware calibrators even when proper scoring rules improve suggests that log-loss and Brier score are more reliable bases for model selection in such settings.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The proof chain assumes that each fitted calibrator is the ECE-optimal member of its parametric family for the cluster it is fit to, but the experiments fit calibrators by log-loss; a direct test would be to fit by ECE and see whether the inequalities survive.
  • The abstract describes a soft mixture of calibrators with hierarchical shrinkage, but the body's algorithm and theory use a hard cluster assignment; the published guarantees should be read as applying to the hard-partition version.
  • If the optimal-fit assumption is relaxed in favor of realistic distributional assumptions, such as cluster-specific label noise, the AUC implication could fail; constructing a small counterexample would settle the true scope of the theorem.
  • The cluster-binning idea could naturally extend to multiclass problems by computing cluster-wise calibration error over class probabilities, and to regression by binning residuals, though the paper does not develop those extensions.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper proposes Clustered Calibration (CCL), a method that clusters samples by a learned representation (SHAP values, leaf indices, etc.) and fits a separate parametric calibrator per cluster, with hierarchical shrinkage toward a global mapping. It also introduces Clustered ECE (CECE), a calibration metric that bins samples by the learned clusters. The paper claims theoretical guarantees: (i) per-cluster ECE of cluster-specific calibrators is upper-bounded by global calibration, (ii) CECE of the ensemble is upper-bounded by unified CECE, and (iii) lower CECE implies higher ROC-AUC (Theorem 2.5), making CECE a model-selection metric that optimizes both calibration and discrimination. Experiments on tabular datasets and image/text benchmarks are presented as supporting these claims.

Significance. If the theoretical guarantees were sound, the paper would make a substantial contribution: a calibration metric that provably selects models with better discrimination as well as calibration is genuinely useful, and the CCL framework is generic and empirically broad. The paper also makes a sensible point that fixed-bin ECE can mis-rank certain calibrators. However, the theoretical backbone is not sound: the main implication from CECE to AUC is invalid, and the auxiliary assumption used in the proofs is not satisfied by the implemented calibrators. The empirical results may suggest CCL is a useful heuristic, but the central claimed guarantee is unsupported.

major comments (4)
  1. [Lemma 2.4 / Theorem 2.5] The proof that lower CECE implies higher AUC is invalid. CECE is a cluster-bin average absolute deviation between mean predicted probability and observed frequency; it imposes no constraint on the relative ordering of individual scores, which is what ROC-AUC measures. A per-cluster calibrator can drive CECE to zero while destroying within-cluster score separation, e.g., by mapping all scores in a cluster to the cluster base rate, making AUC approach 0.5. Step (a)(ii) states 'for a given acceptance threshold th, the accuracy of the accepted samples ... will increase ... Therefore, a higher TPR/FPR and thus a higher AUC.' Accuracy of accepted samples does not imply higher TPR and FPR at every threshold, and higher TPR/FPR pointwise does not imply higher AUC. This flaw is independent of the ECE-optimality assumption in §2.5; even granting that assumption, the implication does not follow. Th
  2. [§2.5 Note] All theoretical results rely on the assumption that the trained parametric calibration method c denotes c*, the model minimizing ECE(c,X). The implemented calibrators—Platt, temperature scaling, beta, and Dirichlet—are fitted by likelihood/log-loss minimization, not by minimizing ECE. Consequently, Lemma 2.1's contradiction argument fails: a global calibrator can have smaller ECE on a cluster than the cluster-specific calibrator, because the cluster-specific calibrator is not optimizing that objective. The upper bounds in Lemmas 2.1 and 2.3 are therefore not consequences of the training procedure. Table 2 illustrates the mismatch: on WiDS, CCL-Platt has CECE 0.0087 but ECE 0.0096, while unified Platt has CECE 0.0117 but ECE 0.0051—the unified version has lower standard ECE. The assumption is not a harmless idealization; it is contradicted by the method's own implementation.
  3. [Lemma 2.3] The claimed upper bound CECE(C,CL,X) ≤ CECE(c,X) is essentially a restatement of the ECE-optimality assumption rather than a proven result. Step 4 of the proof says 'Should there be a better set of parameters, they would have been learned when training c on B_i,' which is precisely the c=c* assumption. Since the cluster-specific calibrators are not ECE-optimal in practice, a global calibrator can outperform the cluster-specific one on a given cluster, and the inequality can fail. The paper does not provide any independent argument for this bound, so it cannot support the subsequent AUC claims.
  4. [§3.1 Experiment Setup] The number of clusters k is selected by the elbow method, and the paper tests partitions of 5 to 100 clusters, but no sensitivity analysis is reported: the chosen k per dataset, representation, and clustering method are not given, nor is stability of the results across k. CECE and the CCL predictions depend directly on the partition, so it is unclear whether the claimed improvements and model-selection rankings in Tables 2–4 are robust or an artifact of a particular elbow choice. This is a load-bearing empirical gap for the model-selection claim.
minor comments (5)
  1. [Figure 2] The caption says 'Claustered Calibration'—typo for 'Clustered Calibration.'
  2. [§3.2.2] The text states a paired t-test was performed for 'the three parametric calibration methods (Platt, Beta, TS)' and later 'for the four parametric calibration methods (Platt, Beta, TS, Dirichlet).' This inconsistency should be corrected.
  3. [Table 4] The row for threshold 0 is confusing: with no rejection, the error rate should be the overall error, but values differ from the full evaluation (e.g., 0.019 vs 0.002 for the base classifier). The threshold semantics need clarification.
  4. [Abstract and §3.1] The abstract says 'six tabular datasets,' but the experiments section lists ten data sources. Please align the counts.
  5. [General] No code or data availability statement is provided. Given the empirical nature of the paper, releasing code would aid reproducibility.

Circularity Check

2 steps flagged · score 6.0 of 10

The CECE guarantee is partly definitional: Lemma 2.1 restates the §2.5 ECE-optimality assumption, and Lemma 2.4 imports the AUC conclusion it claims to prove; Theorem 2.5 is therefore not an independent derivation.

  1. self definitional [Section 2.5 (Note and Lemma 2.1), propagated through Lemma 2.3]
    "Note: In the analysis below, it is assumed that for any given set of samples, X, the trained parametric calibration method, c, denotes c*. That is, the model with the optimal parameter set, θ, which minimizes the calibration error ECE(c,X). ... Contradiction to the training method of c: Should there be a better set of parameters, they would have been learned when training c_i on cl_i."

    Lemma 2.1 asserts ECE(c_i,cl_i) ≤ ECE(c,cl_i) as a 'result' of CCL, but with the Note's definition of c as the ECE-minimizing parameterization on any sample set, this inequality is exactly the definition of per-cluster optimality. The proof does not use the functional form of Platt/TS/Beta/Dirichlet; it merely restates the assumption. Moreover, the implemented methods minimize log-likelihood/Brier-type losses, not ECE, so the claimed upper bound is not a consequence of the actual training procedure.

  2. other [Section 2.6, Lemma 2.4 and Theorem 2.5]
    "As a result, for a given acceptance threshold th, the accuracy of the accepted samples of CCL will increase relative to c, by the definition of ECE. Therefore, a higher TPR/FPR and thus a higher AUC."

    This step assumes the ranking property being proved. CECE is a bin-averaged |mean predicted probability − observed frequency| quantity over fixed cluster bins; it imposes no constraint on the relative order of individual scores, which is what ROC-AUC measures. A cluster-wise calibrator can lower CECE to zero by mapping every score in a cluster to the cluster base rate, destroying score separation and dropping AUC to 0.5. Hence the inference 'lower CECE ⇒ higher AUC' does not follow from the calibration bound; the AUC conclusion is inserted into the proof rather than derived.

full rationale

The paper's theoretical core is not a self-contained first-principles derivation. Claim 1/Lemma 2.1 reduces to the §2.5 assumption that every trained parametric calibrator is the ECE-optimal parameterization; the actual calibrators are fitted by likelihood/log-loss, so the bound is both definitional under the assumption and not satisfied by the experiments. Lemma 2.4 then takes the calibration improvement and asserts an AUC improvement without any linking equation; Theorem 2.5 inherits this unsupported step. This is partial circularity rather than a fully forced result: the empirical comparisons use external NLL, Brier, AUC, and accuracy on public benchmarks, and no load-bearing self-citation chain was found (the only author self-citation, [39], is background on reject options). Those empirical results are independently informative; what is circular/definitional is the claimed guarantee that lower CECE forces higher AUC.

Assumptions & free parameters 5 free parameters · 5 assumptions · 2 invented entities

The central theoretical claims rest on two unproven premises: that fitting calibrators is equivalent to minimizing ECE per cluster, and that within-cluster probability shifts improve AUC. Both are contradicted by standard practice. Cluster counts and cluster assignments are learned from data with no sensitivity analysis. The only genuinely independent content is the empirical comparison, which lacks effect sizes and code.

free parameters (5)
  • Number of clusters k = not reported per dataset; selected by elbow method from 5-100
    k controls clustering granularity and hence which calibrator each sample receives; no sensitivity analysis or fixed rule is reported (§3.1).
  • Cluster assignment model (K-means/agglomerative) = fitted on train+calibration embeddings
    The partition defining CECE bins is learned from the data; results depend on clustering algorithm and seed.
  • Per-cluster calibration parameters = fit per cluster on calibration split
    Standard parameters (Platt A,B; TS temperature; Beta/Dirichlet parameters), but their per-cluster estimates are central to the method.
  • Base classifier hyperparameters = chosen by Optuna
    The XGBoost model and its learned representation/leaf structure determine the clusters; hyperparameter search is part of the pipeline.
  • Rejection thresholds = stepped 0.1 from 0 to 0.9 in Section 4
    Used only in the rejection application; reported as a threshold sweep, not a fitted constant.
assumptions (5)
  • ad hoc to paper Trained calibration functions are ECE-optimal on their training set (c = c*)
    Stated in Note §2.5; false for log-loss-trained calibrators; load-bearing for Lemmas 2.1 and 2.3.
  • domain assumption CECE bin assignment is invariant to the calibrated output
    CECE bins by cluster ID from the representation, not by score; this is by construction (§2.4), enabling fixed-bin comparison.
  • ad hoc to paper Within-cluster shifts of predicted probabilities toward observed frequency raise TPR/FPR at every threshold, hence AUC
    Unstated in Lemma 2.4; violated by monotone calibrators, which preserve rank and therefore AUC.
  • domain assumption Learned representations (SHAP/leaf embeddings) capture calibration-relevant subpopulations
    Empirically motivated (§3.2.1); no guarantee that clusters correspond to calibration-error heterogeneity.
  • domain assumption Base classifier f is fixed before calibration
    Standard practice; CCL trains calibration on precomputed scores, so there is no feedback into f.
invented entities (2)
  • Soft mixture with hierarchical shrinkage (abstract)
    purpose: Blend cluster-specific calibrators toward a global mapping to maintain stability; central to the abstract's description of CCL.
    Mentioned only in the abstract; no equations, algorithm, or experiments in the full text; cannot be evaluated or replicated.
  • Clustered ECE (CECE) metric
    purpose: New calibration metric used for model selection; supposed to rank models by both calibration and AUC.
    Theoretical justification rests on flawed lemmas; no external benchmark establishes its validity as a model-selection guarantee.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Clustered Calibration: Representation-Aware Probability Calibration via Learned Subpopulations." pith.science (2026). https://pith.science/paper/PT7SUAO7

@misc{pith2026251019328,
  author       = {Pith},
  title        = {Pith review of: Clustered Calibration: Representation-Aware Probability Calibration via Learned Subpopulations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PT7SUAO7}},
  note         = {Machine review of arXiv:2510.19328}
}
read the original abstract

Ensuring that predicted probabilities align with observed frequencies is critical in high-stakes domains such as clinical decision support, autonomous driving and financial risk assessment. Existing calibration methods typically apply a single global transformation or rely on post-hoc binning over predicted confidences, limiting their ability to exploit heterogeneous reliability across sub-populations. We propose Clustered Calibration, a representation-aware framework that identifies sub-populations via clustering in learned feature spaces (e.g., coverage vectors, SHAP values, CNN activations, Transformer embeddings) and fits a soft mixture of cluster-specific parametric calibrators under hierarchical shrinkage toward a global mapping. This design yields context-specific calibration while maintaining global stability. Across six tabular datasets and additional image and text benchmarks, clustered calibration consistently improves or matches strong global calibrators in terms of negative log-likelihood and Brier score, while preserving AUC and accuracy. We further show, both analytically and empirically, that fixed-bin Expected Calibration Error (ECE) can mis-rank soft, region-aware calibrators even when proper scoring rules improve, and we advocate for log-loss and Brier as more reliable bases for model selection in such settings.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

40 extracted references · 2 canonical work pages

  1. [1]

    Advances in Neural Information Processing Systems34, 15682–15694 (2021)

    Minderer, M., Djolonga, J., Romijnders, R., Hubis, F., Zhai, X., Houlsby, N., Tran, D., Lucic, M.: Revisiting the calibration of modern neural networks. Advances in Neural Information Processing Systems34, 15682–15694 (2021)

    2021

  2. [2]

    Journal of Neurosurgery: Spine32(6), 985– 987 (2020)

    Staartjes, V.E., Kernbach, J.M.: Importance of calibration assessment in machine learning–based predictive analytics. Journal of Neurosurgery: Spine32(6), 985– 987 (2020)

    2020

  3. [3]

    In: Proceedings of the AAAI Conference on Artificial Intelligence, vol

    Naeini, M.P., Cooper, G., Hauskrecht, M.: Obtaining well calibrated probabilities using bayesian binning. In: Proceedings of the AAAI Conference on Artificial Intelligence, vol. 29 (2015)

    2015

  4. [4]

    arXiv preprint arXiv:1508.05154 (2015) 17

    Nguyen, K., O’Connor, B.: Posterior calibration and exploratory analysis for natural language processing models. arXiv preprint arXiv:1508.05154 (2015) 17

    arXiv 2015

  5. [5]

    Platt, J.: Probabilistic outputs for support vector machines and comparisons to regularized likelihood methods. Adv. Large Margin Classif.10(2000)

    2000

  6. [6]

    In: International Conference on Machine Learning, pp

    Guo, C., Pleiss, G., Sun, Y., Weinberger, K.Q.: On calibration of modern neu- ral networks. In: International Conference on Machine Learning, pp. 1321–1330 (2017). PMLR

    2017

  7. [7]

    Electronic Journal of Statistics11, 5052–5080 (2017)

    Kull, M., Silva Filho, T.M., Flach, P.: Beyond sigmoids: How to obtain well- calibrated probabilities from binary classifiers with beta calibration. Electronic Journal of Statistics11, 5052–5080 (2017)

    2017

  8. [8]

    arXiv preprint arXiv:1910.12656 (2019)

    Kull, M., Perello-Nieto, M., K¨ angsepp, M., Song, H., Flach, P., et al.: Beyond tem- perature scaling: Obtaining well-calibrated multiclass probabilities with dirichlet calibration. arXiv preprint arXiv:1910.12656 (2019)

    arXiv 1910

Show all 40 references
  1. [9]

    In: Icml, vol

    Zadrozny, B., Elkan, C.: Obtaining calibrated probability estimates from decision trees and naive bayesian classifiers. In: Icml, vol. 1, pp. 609–616 (2001)

    2001

  2. [10]

    In: Proceedings of the Eighth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp

    Zadrozny, B., Elkan, C.: Transforming classifier scores into accurate multiclass probability estimates. In: Proceedings of the Eighth ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 694–699 (2002)

    2002

  3. [11]

    Findings of the Association for Computational Linguistics: ACL 2022, 3673–3684 (2022)

    Singh, R., Goshtasbpour, S.: Platt-bin: Efficient posterior calibrated training for nlp classifiers. Findings of the Association for Computational Linguistics: ACL 2022, 3673–3684 (2022)

    2022

  4. [12]

    In: 2022 30th European Signal Processing Conference (EUSIPCO), pp

    Frenkel, L., Goldberger, J.: Network calibration by temperature scaling based on the predicted confidence. In: 2022 30th European Signal Processing Conference (EUSIPCO), pp. 1586–1590 (2022). IEEE

    2022

  5. [13]

    In: 2019 IEEE/CVF International Conference on Computer Vision Workshop (ICCVW), pp

    Ji, B., Jung, H., Yoon, J., Kim, K.,et al.: Bin-wise temperature scaling (bts): Improvement in confidence calibration performance through simple scaling tech- niques. In: 2019 IEEE/CVF International Conference on Computer Vision Workshop (ICCVW), pp. 4190–4196 (2019). IEEE

    2019

  6. [14]

    Advances in Neural Information Processing Systems32(2019)

    Kumar, A., Liang, P.S., Ma, T.: Verified uncertainty calibration. Advances in Neural Information Processing Systems32(2019)

    2019

  7. [15]

    In: International Conference on Machine Learning, pp

    Zhang, J., Kailkhura, B., Han, T.Y.-J.: Mix-n-match: Ensemble and composi- tional methods for uncertainty calibration in deep learning. In: International Conference on Machine Learning, pp. 11117–11128 (2020). PMLR

    2020

  8. [16]

    In: 2021 29th European Signal Processing Conference (EUSIPCO), pp

    Frenkel, L., Goldberger, J.: Network calibration by class-based temperature scal- ing. In: 2021 29th European Signal Processing Conference (EUSIPCO), pp. 1486–1490 (2021). IEEE

    2021

  9. [17]

    In: International Con- ference on Machine Learning, pp

    H´ ebert-Johnson, U., Kim, M., Reingold, O., Rothblum, G.: Multicalibration: 18 Calibration for the (computationally-identifiable) masses. In: International Con- ference on Machine Learning, pp. 1939–1948 (2018). PMLR

    1939

  10. [18]

    In: International Conference on Machine Learning, pp

    Yang, B., Fu, X., Sidiropoulos, N.D., Hong, M.: Towards k-means-friendly spaces: Simultaneous deep learning and clustering. In: International Conference on Machine Learning, pp. 3861–3870 (2017). PMLR

    2017

  11. [19]

    In: International Conference on Machine Learning, pp

    Xie, J., Girshick, R., Farhadi, A.: Unsupervised deep embedding for clustering analysis. In: International Conference on Machine Learning, pp. 478–487 (2016). PMLR

    2016

  12. [20]

    arXiv preprint arXiv:1710.00486 (2017)

    Gopinath, D., Katz, G., Pasareanu, C.S., Barrett, C.: Deepsafe: A data-driven approach for checking adversarial robustness in neural networks. arXiv preprint arXiv:1710.00486 (2017)

    2017 arXiv

  13. [21]

    In: Guyon, I., Luxburg, U.V., Bengio, S., Wallach, H., Fergus, R., Vishwanathan, S., Garnett, R

    Lundberg, S.M., Lee, S.-I.: A unified approach to interpreting model predictions. In: Guyon, I., Luxburg, U.V., Bengio, S., Wallach, H., Fergus, R., Vishwanathan, S., Garnett, R. (eds.) Advances in Neural Information Processing Systems 30, pp. 4765–4774. Curran Associates, Inc...

    2017

  14. [22]

    In: Proceedings of the Eighth International Workshop on Data Mining for Online Advertising, pp

    He, X., Pan, J., Jin, O., Xu, T., Liu, B., Xu, T., Shi, Y., Atallah, A., Herbrich, R., Bowers, S.,et al.: Practical lessons from predicting clicks on ads at facebook. In: Proceedings of the Eighth International Workshop on Data Mining for Online Advertising, pp. 1–9 (2014)

    2014

  15. [23]

    PhysioNet (2020)

    Johnson, A., Bulgarelli, L., Pollard, T., Horng, S., Celi, L.A., Mark, R.: MIMIC- IV (version 0.4). PhysioNet (2020). https://doi.org/10.13026/a3wn-hq05 . https: //physionet.org/content/mimiciv/0.4/

    2020 doi

  16. [24]

    Kaggle (2020)

    Karen, M., Marzyeh, G., Meredith Lee, N., Sharada Kalanidhi, s.: WiDS Datathon 2020. Kaggle (2020). https://kaggle.com/competitions/ widsdatathon2020

    2020

  17. [25]

    UCI Machine Learning Repository

    Clore, J., Cios, K., DeShazo, J., Strack, B.: Diabetes 130-US Hos- pitals for Years 1999-2008. UCI Machine Learning Repository. DOI: https://doi.org/10.24432/C5230J (2014)

    1999 doi

  18. [26]

    UCI Machine Learning Repository

    Becker, B., Kohavi, R.: Adult. UCI Machine Learning Repository. DOI: https://doi.org/10.24432/C5XW20 (1996)

    1996 doi

  19. [27]

    UCI Machine Learning Repository

    Yeh, I.-C.: Default of Credit Card Clients. UCI Machine Learning Repository. DOI: https://doi.org/10.24432/C55S3H (2009)

    2009 doi

  20. [28]

    UCI Machine Learning Repository

    Roe, B.: MiniBooNE particle identification. UCI Machine Learning Repository. DOI: https://doi.org/10.24432/C5QC87 (2005) 19

    2005 doi

  21. [29]

    IEEE Dataport (2023)

    Hassan, A.: Stroke Prediction Dataset. IEEE Dataport (2023). https://doi.org/ 10.21227/mxfb-sc71 . https://dx.doi.org/10.21227/mxfb-sc71

    2023 doi

  22. [30]

    Kaggle (2014)

    Campanelli, P.: League of Legends Ranked Matches. Kaggle (2014). https:// www.kaggle.com/datasets/paololol/league-of-legends-ranked-matches

    2014

  23. [31]

    Grinsztajn, L., Oyallon, E., Varoquaux, G.: Why do tree-based models still out- perform deep learning on typical tabular data? Advances in neural information processing systems35, 507–520 (2022)

    2022

  24. [32]

    In: Proceedings of the 22nd Acm Sigkdd International Conference on Knowledge Discovery and Data Mining, pp

    Chen, T., Guestrin, C.: Xgboost: A scalable tree boosting system. In: Proceedings of the 22nd Acm Sigkdd International Conference on Knowledge Discovery and Data Mining, pp. 785–794 (2016)

    2016

  25. [33]

    In: Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, pp

    Akiba, T., Sano, S., Yanase, T., Ohta, T., Koyama, M.: Optuna: A next- generation hyperparameter optimization framework. In: Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, pp. 2623–2631 (2019)

    2019

  26. [34]

    IEEE transactions on information theory28(2), 129–137 (1982)

    Lloyd, S.: Least squares quantization in pcm. IEEE transactions on information theory28(2), 129–137 (1982)

    1982

  27. [35]

    Journal of the American statistical association58(301), 236–244 (1963)

    Ward Jr, J.H.: Hierarchical grouping to optimize an objective function. Journal of the American statistical association58(301), 236–244 (1963)

    1963

  28. [36]

    Journal of Machine Learning Research9(86), 2579–2605 (2008)

    Maaten, L., Hinton, G.: Visualizing data using t-sne. Journal of Machine Learning Research9(86), 2579–2605 (2008)

    2008

  29. [37]

    In: Proceedings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning (EMNLP-CoNLL), pp

    Rosenberg, A., Hirschberg, J.: V-measure: A conditional entropy-based exter- nal cluster evaluation measure. In: Proceedings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning (EMNLP-CoNLL), pp. 410–420 (2007)

    2007

  30. [38]

    arXiv preprint arXiv:2107.11277 (2021)

    Hendrickx, K., Perini, L., Plas, D., Meert, W., Davis, J.: Machine learning with a reject option: A survey. arXiv preprint arXiv:2107.11277 (2021)

    2021 arXiv

  31. [39]

    In: Artificial Intelligence in Medicine: 20th Inter- national Conference on Artificial Intelligence in Medicine, AIME 2022, Halifax, NS, Canada, June 14–17, 2022, Proceedings, pp

    Salillari, G., Rappoport, N.: Comparison of classification with reject option approaches on mimic-iv dataset. In: Artificial Intelligence in Medicine: 20th Inter- national Conference on Artificial Intelligence in Medicine, AIME 2022, Halifax, NS, Canada, June 14–17, 2022, Proc...

    2022

  32. [40]

    NPJ Digital Medicine4(1), 4 (2021) 20

    Kompa, B., Snoek, J., Beam, A.L.: Second opinion needed: communicating uncertainty in medical machine learning. NPJ Digital Medicine4(1), 4 (2021) 20

    2021

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.