Bias-corrected score decomposition for generalized quantiles

• 成果类型: Article
• 署名作者: Ehm, W.; Ovcharov, E. Y.
• 署名单位: Heidelberg Institute for Theoretical Studies; Bulgarian Academy of Sciences
• 刊物名称: BIOMETRIKA
• ISSN/ISSBN: 0006-3444
• DOI: 10.1093/biomet/asx004
• 发表日期: 2017
• 页码: 473480
• 关键词: forecast verification; reliability; RESOLUTION; FRAMEWORK

摘要

Decompositions of the score of a forecast represent useful tools for assessing its performance. We consider local score decompositions permitting detailed forecast assessments across a spectrum of conditions of interest. We derive corrections to the bias of the decomposition components in the framework of point forecasts of quantile-type functionals, and illustrate their performance by simulation. Related bias corrections have thus far only been known for squared error criteria.