data bug 2023-01-02

[lucasrodes]

Hi @3dgiordano!
I hope everything is going well for you.

I just detected some irregularities for data between 2023-01-02 and 2023-07-31. Ther is a sudden drop in the number of vaccines (first, second, etc. doses too).

You can observe this if you compare the values in Uruguay.csv from 2023-01-01 and the next day.

Thanks!

[3dgiordano]

Thanks @lucasrodes

The origin of the problem is at the source of the Ministry of public health.
The historical data that they are publishing in the open data system is wrong, since the historical published erroneous data from the date you indicate, these were overwritten and that ended up being propagated in the published data.
I reported the problem to the Ministry of public health, I hope they can resolve it soon.

In the meantime, I am going to see how to disable the import of the history, leave only the online information active and try to recover the lost information.

Thank you very much for reporting the problem.

[3dgiordano]

Hi @lucasrodes

The Ministry of Public Health resolved the problems in its open data.

It is worth clarifying that they have not yet resolved some problems that exist regarding the +5 dosis information in their historical sources.

I am still waiting for the Ministry to resolve the problems in its data so that I can have a complete view.
Currently, what is certain about what is reported are the first dose and first booster vaccinations, first dose and second, but not the fifth dose and more.
These fifth doses and more already represent more than 184 thousand doses of difference and are those quantities where a fluctuation can be seen. The "online" system record the full data, but the historic batch override that data (and some days that data not it's overwritten by historic and in that place the +184 thousand and the next day the -184thousand appears.

Outside of that detail, all data is now back online in open data and it reports all data for doses less than the fifth dose and more (like before the reported problem on general data lost in historic information from MSP).

Unfortunately, I do not see correct data reflected in OWID since December 24 2022.
Is there any problem in data processing?