Seminar: An Inside Look at the CDC's Covid-19 Forecasting

A PDF of the presentation is available here: CS Seminar 5 - Covid Presentation Materials.

 Speaker

Steve McConnell has been a leading expert in software estimation, software measurement, and software engineering practices for more than 20 years. Steve is best known as the author of the software industry classics Code Complete, Software Estimation: Demystifying the Black Art, and More Effective Agile. Steve is the CEO of Construx Software, and he is currently contributing to the CDC's Covid-19 forecast "Ensemble" model.

Abstract

Rarely has there been a situation in which forecasting a new phenomenon is so critically important. Lives depend on knowing where Covid-19 will strike next, and billions in economic activity also ride on accurately predicting the course of the pandemic. In this context, the Centers for Disease Control and Prevention (CDC) has adopted an open source, community-based approach to building its Ensemble forecasting model. The model forecasts cases and deaths at both national and state levels for periods of 1-4 weeks. Contributors to the Ensemble model include groups from MIT, Harvard, Johns Hopkins, USC, Los Alamos National Labs, and Seattle University alum Steve McConnell. In this talk, Steve will overview the different forecasting methods of the different groups, share data on forecast accuracy at the national and state levels, and discuss possibilities and limitations of forecasting the Covid-19 pandemic.

Visit the CS Seminars Page for more information about Computer Science Seminars