software . 2021

MIT_Crit_Data-GBCF: A Non-Mechanistic Approach to Real-Time Forecasting of US COVID-19 Mortality Using Machine Learning and Novel Digital Data

Sundar, Saketh;
Open Access
  • Published: 09 Oct 2021
  • Publisher: Zenodo
About the model This model uses a gradient boosted regressor with hyperparameter optimization that uses prior COVID cases and deaths as well as demographic, socioeconomic, mobility, and healthcare related county-level covariates. This model forecasts COVID deaths at the county level and aggregates to the state and national level. This model does not directly account for state reopenings and closures but indirectly via changes in the mobility measures. Evaluation of model predictions is done using COVID-19 Forecast Hub Evaluation modules. Our forecasts are featured on the CDC Website. Please visit for more informat...
  • COVID-19
Any information missing or wrong?Report an Issue