Matches in SemOpenAlex for { <https://semopenalex.org/work/W4226109939> ?p ?o ?g. }
Showing items 1 to 59 of
59
with 100 items per page.
- W4226109939 abstract "ML-based predictions are used to inform consequential decisions about individuals. How should we use predictions (e.g., risk of heart attack) to inform downstream binary classification decisions (e.g., undergoing a medical procedure)? When the risk estimates are perfectly calibrated, the answer is well understood: a classification problem's cost structure induces an optimal treatment threshold $j^{star}$. In practice, however, some amount of miscalibration is unavoidable, raising a fundamental question: how should one use potentially miscalibrated predictions to inform binary decisions? We formalize a natural (distribution-free) solution concept: given anticipated miscalibration of $alpha$, we propose using the threshold $j$ that minimizes the worst-case regret over all $alpha$-miscalibrated predictors, where the regret is the difference in clinical utility between using the threshold in question and using the optimal threshold in hindsight. We provide closed form expressions for $j$ when miscalibration is measured using both expected and maximum calibration error, which reveal that it indeed differs from $j^{star}$ (the optimal threshold under perfect calibration). We validate our theoretical findings on real data, demonstrating that there are natural cases in which making decisions using $j$ improves the clinical utility." @default.
- W4226109939 created "2022-05-05" @default.
- W4226109939 creator A5037714222 @default.
- W4226109939 creator A5057788395 @default.
- W4226109939 date "2022-03-18" @default.
- W4226109939 modified "2023-10-14" @default.
- W4226109939 title "Decision-Making under Miscalibration" @default.
- W4226109939 doi "https://doi.org/10.48550/arxiv.2203.09852" @default.
- W4226109939 hasPublicationYear "2022" @default.
- W4226109939 type Work @default.
- W4226109939 citedByCount "0" @default.
- W4226109939 crossrefType "posted-content" @default.
- W4226109939 hasAuthorship W4226109939A5037714222 @default.
- W4226109939 hasAuthorship W4226109939A5057788395 @default.
- W4226109939 hasBestOaLocation W42261099391 @default.
- W4226109939 hasConcept C10347200 @default.
- W4226109939 hasConcept C105795698 @default.
- W4226109939 hasConcept C119857082 @default.
- W4226109939 hasConcept C149782125 @default.
- W4226109939 hasConcept C154945302 @default.
- W4226109939 hasConcept C15744967 @default.
- W4226109939 hasConcept C165838908 @default.
- W4226109939 hasConcept C2776502983 @default.
- W4226109939 hasConcept C33923547 @default.
- W4226109939 hasConcept C41008148 @default.
- W4226109939 hasConcept C48372109 @default.
- W4226109939 hasConcept C50817715 @default.
- W4226109939 hasConcept C77805123 @default.
- W4226109939 hasConcept C94375191 @default.
- W4226109939 hasConceptScore W4226109939C10347200 @default.
- W4226109939 hasConceptScore W4226109939C105795698 @default.
- W4226109939 hasConceptScore W4226109939C119857082 @default.
- W4226109939 hasConceptScore W4226109939C149782125 @default.
- W4226109939 hasConceptScore W4226109939C154945302 @default.
- W4226109939 hasConceptScore W4226109939C15744967 @default.
- W4226109939 hasConceptScore W4226109939C165838908 @default.
- W4226109939 hasConceptScore W4226109939C2776502983 @default.
- W4226109939 hasConceptScore W4226109939C33923547 @default.
- W4226109939 hasConceptScore W4226109939C41008148 @default.
- W4226109939 hasConceptScore W4226109939C48372109 @default.
- W4226109939 hasConceptScore W4226109939C50817715 @default.
- W4226109939 hasConceptScore W4226109939C77805123 @default.
- W4226109939 hasConceptScore W4226109939C94375191 @default.
- W4226109939 hasLocation W42261099391 @default.
- W4226109939 hasOpenAccess W4226109939 @default.
- W4226109939 hasPrimaryLocation W42261099391 @default.
- W4226109939 hasRelatedWork W1568674531 @default.
- W4226109939 hasRelatedWork W2590078629 @default.
- W4226109939 hasRelatedWork W2951462676 @default.
- W4226109939 hasRelatedWork W2963186828 @default.
- W4226109939 hasRelatedWork W2963699425 @default.
- W4226109939 hasRelatedWork W3190783331 @default.
- W4226109939 hasRelatedWork W4289125447 @default.
- W4226109939 hasRelatedWork W4323323006 @default.
- W4226109939 hasRelatedWork W2185779452 @default.
- W4226109939 hasRelatedWork W2998446700 @default.
- W4226109939 isParatext "false" @default.
- W4226109939 isRetracted "false" @default.
- W4226109939 workType "article" @default.