Comparing trained and untrained probabilistic ensemble forecasts of COVID-19 cases and deaths in the United States
International Journal of Forecasting · 2023-07 · 39 (3) · 1366-1383
DOI: 10.1016/j.ijforecast.2022.06.005 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Forecasting, 2023-07. Verfasst von Evan L Ray und 14 weiteren Autor:innen. Beteiligte Einrichtung: School of Public Health and Health Sciences, University of Massachusetts Amherst, United States of America. Die Arbeit wurde bisher 83-mal zitiert (Crossref) und nennt 55 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: NIGMS NIH HHS, NCIRD CDC HHS, Google Inc.
Abstract (Originalsprache)
The U.S. COVID-19 Forecast Hub aggregates forecasts of the short-term burden of COVID-19 in the United States from many contributing teams. We study methods for building an ensemble that combines forecasts from these teams. These experiments have informed the ensemble methods used by the Hub. To be most useful to policymakers, ensemble forecasts must have stable performance in the presence of two key characteristics of the component forecasts: (1) occasional misalignment with the reported data, and (2) instability in the relative performance of component forecasters over time. Our results indicate that in the presence of these challenges, an untrained and robust approach to ensembling using an equally weighted median of all component forecasts is a good choice to support public health decision-makers. In settings where some contributing forecasters have a stable record of good performance, trained ensembles that give those forecasters higher weight can also be helpful.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- School of Public Health and Health Sciences, University of Massachusetts Amherst, United States of America.
- Machine Learning Department, Carnegie Mellon University, United States of America.
- Department of Data Sciences and Operations, University of Southern California, United States of America.
- COVID-19 Response, U.S. Centers for Disease Control and Prevention, United States of America.
- London School of Hygiene & Tropical Medicine, United Kingdom.
- Chair of Statistical Methods and Econometrics, Karlsruhe Institute of Technology, Germany.
Förderung
- NIGMS NIH HHS
- NCIRD CDC HHS
- Google Inc
- Helmholtz Zentrum Berlin
- McCune Foundation
- Carnegie Mellon University Center for Machine Learning and Health
- National Institutes of Health
Zitieren
APA
Ray, E. L., Brooks, L. C., Bien, J., Biggerstaff, M., Bosse, N. I., Bracher, J., Cramer, E. Y., Funk, S., Gerding, A., Johansson, M. A., Rumack, A., Wang, Y., Zorn, M., Tibshirani, R. J., & Reich, N. G. (2023). Comparing trained and untrained probabilistic ensemble forecasts of COVID-19 cases and deaths in the United States. International Journal of Forecasting, 39(3), 1366-1383. https://doi.org/10.1016/j.ijforecast.2022.06.005
Vancouver
Ray EL, Brooks LC, Bien J, Biggerstaff M, Bosse NI, Bracher J, et al. Comparing trained and untrained probabilistic ensemble forecasts of COVID-19 cases and deaths in the United States. International Journal of Forecasting. 2023;39(3):1366-1383. doi:10.1016/j.ijforecast.2022.06.005
BibTeX
@article{Ray2023,
author = {Ray, Evan L and Brooks, Logan C and Bien, Jacob and Biggerstaff, Matthew and Bosse, Nikos I and Bracher, Johannes and Cramer, Estee Y and Funk, Sebastian and Gerding, Aaron and Johansson, Michael A and Rumack, Aaron and Wang, Yijin and Zorn, Martha and Tibshirani, Ryan J and Reich, Nicholas G},
title = {Comparing trained and untrained probabilistic ensemble forecasts of COVID-19 cases and deaths in the United States},
journal = {International Journal of Forecasting},
year = {2023},
volume = {39},
number = {3},
pages = {1366--1383},
doi = {10.1016/j.ijforecast.2022.06.005}
}
EUVIMED