Here, you will find the raw data for our covid19-projections.com projections, ordered by the date the projections were generated. Under each date directory, you will find a .csv file for each individual state/region/country. You can also look under our combined
directory to see all of our daily projections combined in a single file. The directories are further subdivided by global countries and subregions (US counties and Canadian provinces).
actual_deaths
- Daily deaths as reported by Johns Hopkins University CSSEtotal_deaths
- Cumulative deaths as reported by Johns Hopkins University CSSEpredicted_deaths_mean
,predicted_deaths_lower
,predicted_deaths_upper
- Mean, 2.5th, and 97.5th percentile of our daily deaths projectionspredicted_total_deaths_mean
,predicted_total_deaths_lower
,predicted_total_deaths_upper
(Since 2020-04-02) - Mean, 2.5th, and 97.5th percentile of our cumulative deaths projectionspredicted_new_infected_mean
,predicted_new_infected_lower
,predicted_new_infected_upper
(Since 2020-04-08) - Mean, 2.5th, and 97.5th percentile of our estimates for the total number of newly infected individuals on that day. This is based on the date an individual is first exposed. Note that these estimates include all infections, not just those who tested positive.predicted_current_infected_mean
,predicted_current_infected_lower
,predicted_current_infected_upper
(Since 2020-04-08) - Mean, 2.5th, and 97.5th percentile of our estimates for the total number of individuals that are infected on that day. This is the sum of the past 15 days of newly infected individualspredicted_total_infected_mean
,predicted_total_infected_lower
,predicted_total_infected_upper
(Since 2020-04-08) - Mean, 2.5th, and 97.5th percentile of our estimates for the total number of individuals that have ever been infected. This is the cumulative sum of the newly infected individualsr_values_mean
(since 2020-05-15),r_values_lower
(since 2020-07-30),r_values_upper
(since 2020-07-30) - Mean, lower and upper bound estimates of the Rt value on that day. Starting on 2020-08-24, we began including Rt estimates for the future as well. Note that our Rt lower and upper bounds tend to be on the narrow side.
-
Projections in this repository are rounded to the nearest integer, and hence is missing decimal precision. To avoid rounding errors, we recommend using the cumulative deaths to perform computations on incident deaths (e.g. weekly incident deaths).
-
Our estimates from early April do not incorporate reopenings and probable deaths, two factors that significantly increases the estimates. Furthermore, our April 1-7 projections only go through June 30. Starting on April 8, our projections were extended to August 4.
-
Our first projections on April 1 do not have an upper/lower bound.
-
We added estimates of infected individuals starting on April 8.
-
We added estimates of the R_t value starting on May 15.
-
We send projections to the CDC every Monday using Sunday's data. For example, April 26, May 3, May 10, May 17, etc. As a result, we recommend using those projections for evaluation, since those have the highest impact and were the most carefully vetted. See our open source evaluation.
2020-10-05
- Final model update (using data from 2020-10-04). For more information, read Youyang Gu's blog post. Follow @youyanggu on Twitter for continued COVID-19 insights. Thank you for your support over the past year.
2020-07-22
- We released a major update that tries to better account for increases in cases and deaths from the reopening. See Update Notes on Twitter
2020-07-08
- Extended projection end date from October 1 to November 1
2020-06-23
- We have open-sourced the underlying SEIR simulator behind our model
2020-06-16
- We have open-sourced our code to evaluate COVID-19 models. The goal of this project is to evaluate various models’ historical point forecasts in a transparent, rigorous, and non-biased manner
2020-06-15
- Switch main data source from Johns Hopkins Daily Reports to Johns Hopkins Time Series Summary
2020-06-11
- Extended projection end date from September 1 to October 1
2020-05-26
- Add 7 new countries (Australia, Belarus, Bolivia, Cuba, Honduras, Kuwait, UAE), 2 Canadian provinces (Alberta, British Columbia), and 20 US counties
2020-05-25
- Increase projected end date from August 4 to September 1
2020-05-19
- Add subregion projections (14 US counties + Ontario/Quebec)
2020-05-16
- Add estimate for the effective reproduction value (R_t) over time
2020-05-12
- Add projections for 23 additional countries: Algeria, Argentina, Bangladesh, Chile, Colombia, Dominican Republic, Ecuador, Egypt, Iceland, Israel, Japan, Malaysia, Moldova, Morocco, Nigeria, Pakistan, Panama, Peru, Saudi Arabia, Serbia, South Africa, South Korea, Ukraine
2020-04-24
- Forecasts added to the CDC website
2020-04-23
- Incorporate probable deaths into projections, following updated CDC guidelines
2020-04-20
- First projections submitted to the Centers for Disease Control and Prevention (CDC).
2020-04-15
- Incorporate the relaxing of social distancing in June (see our Assumptions page)
2020-04-13
- Switch main data source from COVID Tracking Project to Johns Hopkins CSSE
2020-04-12
- Add Norway and Russia to projections
2020-04-09
- Increase projected end date from June 30 to August 4
- Add plots for the number of infected individuals
2020-04-07
- Add projections for all European Union countries and 7 additional countries: Brazil, Canada, India, Indonesia, Mexico, Philippines, Turkey
2020-04-05
- Launch covid19-projections.com
2020-04-04
- Separate global data from US data
2020-04-03
- Add 9 international countries for projections: Belgium, France, Germany, Iran, Italy, Netherlands, Spain, Switzerland, United Kingdom
2020-04-02
- Add lower and upper bounds to projections; also project date of peak deaths
- Incorporate international data and add projections for Italy
2020-04-01
- Incorporate US states data and add projections for every state
2020-03-31
- Add first projections for New York and California
- Initial projection end date is June 30
2020-03-30
- Begin project