Friday 13th April 2007, 14.00-16.00PM
Laboratoire d'Océanographie et du Climat: Expérimentation et Approches Numèriques
Meeting room nr. 417, corridor 45-55, 4th floor
4 place de Jussieu
75252 PARIS CEDEX 05

Seminar: Creating probability forecasts of binary events from ensemble predictions and prior information - A comparison of methods.
by Cristina PRIMO, University of Reading

Abstract

Probabilistic forecasting provides the probability that an event of interest occurs. Ensembles of forecasts can be used to estimate that probability. However, prior information may also be available from past data, from expert opinion, or both. Bayesian methods can be used to combine prior information with ensemble data to create probability forecasts of binary events. Several such Bayesian methods will be compared to each other, and to methods such as logistic regression that do not use a Bayesian framework, and illustrated using rainfall forecasts for Reading (UK).

For more informations: C. PRIMO (C. Primo email)

Link to presentation: ppt file

up arrow