Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
A Bayesian Bradley-Terry model to compare multiple ML algorithms on multiple data sets
Jacques Wainer
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| remove xgb results on first two data sets (missing data) (ablation) | BBT · 5 classifiers on 20 data sets | mean P(xgb > svm): 0.56 -> 0.53 | No clear effect |
| remove xgb results on first two data sets (ablation) | BBT · 5 classifiers on 20 data sets | mean P(xgb > dt): 0.83 -> 0.81 | No clear effect |
| swap lognormal 0.5 hyper-prior for Cauchy or normal alternatives (ablation) | BBT · ss and mm use cases | WAIC: 40.56506 vs 40.85225 | No clear effect |
| feed BBT all 16 algorithms instead of only 3 (ablation) | BBT strong interpretation · random 20 data sets | mean P(a>b): mean difference 0.006 | No clear effect |
| feed BBT only 3 algorithms instead of all 16 (ablation) | BBT weak interpretation · random 20 data sets | above.50: smaller above.50 with 3 algorithms | Hurt |
| treat ties as half-victories rounded up for both algorithms (spread policy) | BBT · 5 classifiers on 20 data sets | win/loss counts: dt vs lgbm 0 17 -> 2 19 | Improved |
| replace raw mean difference by local ROPE with Cohen d threshold 0.4 | BBT · cmc data set lgbm vs xgb | win vs tie: difference 0.001 vs sd 0.02 and 0.03 | Improved |
Closest papers by meaning
- Generalized Bradley-Terry Models for Score Estimation from Paired Comparisons (AAAI 2024 · similarity 0.88)
- Pairwise Comparisons without Stochastic Transitivity: Model, Theory and Applications (JMLR 2026 · similarity 0.85)
- Time for a Change: a Tutorial for Comparing Multiple Classifiers Through Bayesian Analysis (JMLR 2017 · similarity 0.85)
- Bivariate Causal Discovery using Bayesian Model Selection (ICML 2024 · similarity 0.83)
- Challenges and Considerations in the Evaluation of Bayesian Causal Discovery (ICML 2024 · similarity 0.82)
- Recursive PAC-Bayes: A Frequentist Approach to Sequential Prior Updates with No Information Loss (NeurIPS 2024 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
