Research map / Learning theory and algorithms / Prediction and decision losses
What Is the Optimal Ranking Score Between Precision and Recall? We Can Always Find It and It Is Rarely F1
Sébastien Piérard, Adrien Deliège, Marc Van Droogenbroeck
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 |
|---|---|---|---|
| use F1 instead of the optimal beta for ranking on fixed-prior distributions | F1 ranking · Uniform distribution, fixed class priors | degree of optimality O: O in [50%, 100%], optimal only for pi+ ~ 0.381 | Hurt |
| use F1 on uniform distribution over all performances | F1 ranking · Uniform distribution over all performances | degree of optimality O: O = 100% (optimal) | No clear effect |
| use SIVF instead of F1 for ranking | SIVF ranking · CDnet 2014 (53 real sets) | mean degree of optimality O: 78.44% -> 56.71% | Hurt |
| use heuristic beta^2 = E[PFP]/E[PFN] instead of F1 | F-beta ranking · CDnet 2014 (53 real sets) | mean degree of optimality O: 78.44% -> 89.53% | Improved |
| use closed-form median-based optimal beta instead of fixed F1 | F-beta ranking · CDnet 2014 (53 real sets) | degree of optimality O: 78% (F1) -> 100% | Improved |
| use F2 as in CADA-RRE challenge to emphasize recall | CADA-RRE ranking · CADA-RRE (16 performances) | rank mimicry: beta=2 vs beta=0.914 for 4x recall weighting by ranks | Hurt |
Closest papers by meaning
- Foundations of the Theory of Performance-Based Ranking (CVPR 2025 · similarity 0.85)
- On ranking via sorting by estimated expected utility (NeurIPS 2020 · similarity 0.83)
- RD-Suite: A Benchmark for Ranking Distillation (NeurIPS 2023 · similarity 0.82)
- Principled Algorithms for Optimizing Generalized Metrics in Binary Classification (ICML 2025 · similarity 0.81)
- Probably Approximately Precision and Recall Learning (NeurIPS 2025 · similarity 0.81)
- A Closer Look at AUROC and AUPRC under Class Imbalance (NeurIPS 2024 · similarity 0.81)
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.
