articles/iso/FRA
utterances-bot opened this issue · 6 comments
France • COVID-19 Data Hub
The deaths for France seem to begin abruptly with a value of 6500 deaths on 03Apr2020, which is not consistent I think with the JHU CSSE data.
Also the recovered seem to be very low (~300k) compared to actual cases (~6million) while worldometers.info reports ~4million.
Hi @Inglezos , thanks for your message.
Regarding the value of 6500 deaths on 03Apr2020 it seems correct and close to JHU CSSE or other sources. This is the first date for which the number of deaths is available from the source we are using for France: https://github.com/opencovid19-fr/data I see no problem with the data here, they are simply missing (and values before this date would be quite unreliable in any case I guess)
Regarding the number of recovered people, yes it very misleading. For France this is the number of recoveries from the hospital only. From the repo above:
total cumulé du nombre de personnes guéries (sorties de l'hôpital)
I'm wondering if there is another source with the time series of recoveries similar to what worldometers is providing.
For the recoveries I am not sure about what is the source of worldometers, I cannot find such a source reliably anywhere.
But regarding the deaths, JHU CSSE in its github reports otherwise. If you navigate to https://github.com/CSSEGISandData/COVID-19/blob/master/csse_covid_19_data/csse_covid_19_time_series/time_series_covid19_deaths_global.csv
and inspect manually the death values for France, you can see the following:
and if you move to the right area you can see that the deaths actually begin from 1 and increment more naturally up to ~6500 on 03Apr2020, not abruptly:
In rawdata-1.csv from https://storage.covid19datahub.io/rawdata-1.zip,
you have though:
while all the previous dates have missing death cases.
Could something have gone wrong during the aggregation you applied to the CSSE data or am I doing something wrong?
Yes thanks for double-checking. I can see that JHU has some data also before 2020-04 but we don't take data from there for France. For France we pull the data from Santé publique France. They don't provide the total number of deaths before 2020-04. Indeed, France provides the number of fatalities in hospital and in elderly homes (EHPAD and ESMS) separately. The total count is the sum of the two. Before 2020-04 no data for EHPAD and ESMS is provided so that no total count is provided as well. I can also see that JHU seems to be off of 1.000 deaths as of today, both with respect to Santé publique France, worldometer, and a quick google search.
I don't think I'm going to fill those data as it seems to be quite unclear where they are coming from. Hope this makes sense!
Yes I agree it makes sense, thank you very much for the investigation and your input on the subject!