0 deaths now appearing in some municipalities after data update
Closed this issue · 4 comments
Aftern an update between March 17th and March 18th, 2021, cases-brazil-cities-time.csv started showing 0 deaths for some municipalities, in dates when deaths had previously been registered. I list specific cases and more details below.
Brasil.io & WC divergence: Brasil.io has been registering 0 deaths for a streak of several days for some municipalities in 2021 (e.g. Campo Grande (MS), Aracaju, Natal, and Belo Horizonte) and for Cuiabá in 2020 (see orange dashed line in graphs below). As of March 17th, 2021, cases-brazil-cities-time.csv did not have these zeroes, as shown by solid yellow line, but data downloaded on/after March 18th, 2021 now contains zeroes for Aracaju, including in the website visualization by municipality (e.g. see Aracaju's line for new deaths now zeroed starting on late February)
Aracaju: deaths as shown on website on 20/03/2021.
Same municipalities, comparing WC data downloaded on 20/03/2021 and Brasil.io.
Do you see any reason for backward revision and zeroing of deaths data?
Hi!
Thank you very much for sharing these plots!
I am aware of these problems with Brasil.IO data. In some states, such as MT (e.g. Cuiabá) and MG (e.g. Belo Horizonte), I had to use data from Ministério da Saúde to avoid all the zeros they have (see #65). In MT, all data from August 2020 are from Ministério da Saúde. In MG, I am using it since mid Feb 2021.
Currently, I am only able to keep using Brasil.IO data up to some date point, and then use only the Ministério da Saúde data. I am already finishing the full integration between these two data sources and I hope to give an update soon.
In any case, I will do the same I did for MT and MG for SE and RN as well in the meantime.
Muito obrigada!
Oi, Wesley,
Este mesmo problema reapareceu, agora para Porto Velho (RO):
Possível que tenham que mudar a fonte novamente. Obrigada!