absorb.md

Benedict Evans

Chronological feed of everything captured from Benedict Evans.

paper / benedictevans / Jun 12

Relative Belief Inference for Uncertainty Quantification in Poisson Models

This work explores the application of relative belief inference for uncertainty quantification, particularly within Poisson models featuring signal and background noise. It establishes that probability theory offers a clear definition of evidence, directly benefiting Bayesian analyses. Relative belief inferences are shown to adhere to Bayesian principles while also meeting frequentist standards like confidence levels when errors are managed. The paper specifically contrasts these relative belief intervals with the established Feldman-Cousins intervals for the described problem.

statistical-inferencebayesian-analysisuncertainty-quantificationparticle-physicsconfidence-intervalsdata-analysis
“Probability theory provides a definitive measure of evidence for or against an event, applicable in Bayesian analysis and imposing constraints on inferences.”