Adaptively stacked ensembles for influenza forecasting with incomplete data
Date
May 20, 2019 12:00 AM
Event
Modeling of Infectious Disease Agent Study
Location
Bethesda, Maryland
Influenza prediction can better prepare public health officials for seasonal outbreaks. We propose an adaptive multi-model ensemble forecast method that changes model weights week-by-week throughout the flu season. Without historical training data or when models do not have a track-record, an adaptive model can enhance the public health impact of ensemble forecasts.
tom mcandrew
Assistant Professor
I am a computational scientist with a methodological focus on developing ensemble forecasting algorithms and extracting statistical information from unstructured human judgment data. The areas of application that interest me most are building tools to combine forecasting and predictive models in the health sciences.