HYDRA – Pacific Northwest National Laboratory
HYDRA is a statistical framework for designing an aggregate forecasting model that is less susceptible to variability. It iteratively tunes, augments, and then combines the strengths from an ensemble of competing methods to generate a single model that is more accurate and reliable than any ensemble constituent. HYDRA’s possible applications range from predicting short- and long-term energy needs in the power grid (one industry collaborator reduced forecasting errors by 65% and is estimated to save up to $100 million annually) to detecting early indications of disease in cattle (reducing the typical $5 billion annual loss). Energy I-Corps will help this team to more clearly define and demonstrate HYDRA’s potential applications and transform it into a high-impact technology.