朝夕说 · 英语阅读

Evaluation of short-term multi-target respiratory forecasts over winter 2024-25 in England using sub-ensemble contribution analyses

C2科学420 词约 2 分钟

BackgroundEpidemic forecasting research often assesses ensembles and their component models using probabilistic scoring rules. Quantifying how individual models affect ensemble performance is challenging, particularly across multiple targets and spatial scales.

MethodsWe present Winter 2024-25 forecasts of Influenza and COVID-19 hospital admissions in England and conduct a retrospective simulation using the operational component models. Forecasts were scored using the per capita weighted interval score (pcWIS) for counts and the ranked probability score (RPS) for ordinal trend direction. We compared operational retrospective forecasts, used generalised additive models (GAMs) to estimate the expected change in score from the inclusion of a model in a sub-ensemble (an ensemble formed from a subset of available models), and used Pareto analysis to understand which sub-ensembles were Pareto-optimal across scoring rules.

Kennedy, J. C. et al. · CC-BY 4.0

朝夕说 · 听说读写背单词 · 赣ICP备2026010754号

免费继续阅读全文 · 查词 · AI 精讲