Data Issue: Dates with negative case numbers and deaths.
Jonathan-Nyquist opened this issue · comments
Describe the issue:
- [] Incorrect number of cases or deaths
- [X ] Suspicious number of cases or deaths
- Missing data for a locality
- Missing time-series information
- Other
Fuller details
In the data set rolling-averages/us.csv there are days with a negative number of cases, and what is more disturbing: negative deaths (Reincarnation? Covid induced zombies?)
@Jonathan-Nyquist Thanks for the clear example here. What you're seeing is probably the result of a downward-revised count of deaths from one or more of our sources, which often happens when local health departments or other officials realized they had duplication or other death-certification challenges. Our README that lives alongside our rolling-average files has a deeper explanation of our methodology and how we judge anomolous entries in our data set and its calculated figures.
Identified anomalies are often because of officials making revisions to improve the overall quality of the data they have released. Many small anomalies due to backlogs of cases or minor revisions of previously announced numbers are not included here, particularly at the county level. There are no listed anomalies from very early in the pandemic. When deciding whether to list an anomaly, we judge whether a member of the public would need that note to understand and put in context that day’s case or death count.
You can search our list of anomalies here in GitHub for the dates in your example, using its filter mechanism.
