US State level recoveries to be removed

[CSSEGISandData]

On the evening of March 7, 2021, we will remove all U.S. state-level “recovered” and "active cases" data from the the JHU CSSE COVID-19 Dashboard and associated data files posted in our Git repository. We've determined that most “recovered” metrics are estimates rather than precise figures, and therefore may be incomplete or lacking in quality. We will no longer be reporting these metrics.

[niyack]

I get why you decided to remove the recovered counts but the active cases actually has value. Why would you take active cases out????

[ddesroches]

Since active cases is "cumulative minus recovered", it has the same imprecision as "recovered" so makes sense to remove it.

[niyack]

but now you are not painting the true sense of where we are with the virus. also you have had active since the beginning and without the recovery numbers for some states, why now it is an issue for Active counts?

I do not mean to come off harsh, it is just that the active numbers tells us where we are really at with the virus. the confirmed counts do not help to paint that picture. the active counts is also what separates your data source from the others..

[arbitar]

@CSSEGISandData While the sharing of your data is definitely appreciated, I believe a more helpful response to discovering that a subset of your data may not be as precise as the rest (note: not 100% imprecise) is to separate it and denote it as less-reliable from the precise data, not eliminate the data entirely. These 'estimates', after all, are presumably being generated by some other means other than random chance, and still may have statistical significance.

I propose the creation of "Recovered_Estimate", "Active_Estimate" or "Recovered_Reported", "Active_Reported" columns to the affected data, which can remain empty for all data sets with precise numbers, but populated for the affected US ones. This forces consumers of your data to recognize that the data may be estimated, but still allows people to draw inferences and trends from it, which is extremely useful!

[bhaden94]

@CSSEGISandData Can we expect the same change for country data in the future?