COVID-19 Projections Using Machine Learning

The main repository for covid19-projections.com, created by Youyang Gu. This contains a comprehensive library of not only the most recent updates but also all historical updates. If you are only interested in the last infections estimates made in March 2021, click here. See a list of all of our COVID-19 repositories below.

Warning: If you plan to clone this repository, please be aware that this repository takes approximately 4GB of space. This is because it contains all of our historical estimates and projections, not just the latest ones.

2021-03-08 Update: We made our last daily update on March 7. This repository will no longer be maintained. Read Youyang Gu's One Year Later blog post for a detailed explanation and for a list of alternate resources.

2020-11-19 Update: We have uploaded our raw US infection estimates in the infection_estimates directory. For a slimmed down version that only has the latest infection estimates, visit this repository.

2020-10-06 Update: Projections dated 2020-10-04 was our last model update. For more information, read Youyang Gu's blog post. Follow @youyanggu on Twitter for continued COVID-19 insights.

We take a data-driven approach rooted in epidemiology to forecast infections and deaths for the COVID-19 / coronavirus epidemic in the US and around the world.

Description of Folders

You can click on each folder to view the README for that folder.

• c19pro_score - Raw data for our C19Pro Score
• implied_ifr - Implied Infection Fatality Rate (IIFR) estimates
• infection_estimates - Infection estimates for the US
• projections - Raw daily projections generated by our model
• r_values - Reproduction number (R) estimates
• reich_forecasts - Reich forecasts follow the format specified by the Reich Lab
• tests_target - Testing targets for every region based on our forecasts

The underlying SEIR simulator used to generate these projections is open source: https://github.com/youyanggu/yyg-seir-simulator. The simulator repository also includes the best learned values of the various parameters for every region.

Our projections are featured on the CDC website.

We have also open-sourced our evaluation of our model and other CDC models: https://github.com/youyanggu/covid19-forecast-hub-evaluation.

Other Repositories

In addition to our main repository here, we have several other COVID-19-related repositories:

• Infections Estimates
• Historical CDC Vaccination Data
• COVID-19 Datasets
• SEIR Simulator
• Model Evaluations

Questions? Contact me on Twitter at @youyanggu.

covid19-projections.com is powered by GitHub Pages.