We compare performance of major COVID-19 models to the Caltech CS156 ensemble model in forecasting the pandemic. Starting from May 10, the CS156 models that had performed best in prior weeks were ensembled with equal voting.
As a benchmark, we compare to IHME projections made on May 9, as well as other notable models that released predictions between May 9 and May 12. Thus, models marked "5/11" or later have an advantage over the CS156 model.
Data is reported by the New York Times. The errors are computed statewide, and a perfect model corresponds to an RMSE of zero.
CS156 RMSE: 17.9
IHME RMSE: 22.1