Confirmed cases in France on 04-14 are less than on 04-13

[LoretoSanchez]

I just found that confirmed cases in France on 04-14 are about 5000 cases less than on 04-13.

[gbhgit]

hi, i found the same problem

[PavelGorbanj]

it's a duplicate that was already discussed in #2226

[Aureuum]

@LoretoSanchez Yes, the time series data for France since 4/4 are in large excess (about 20k-30k) compared to the correct official data. The probable cause is that the jhu.edu site itself shows the wrong data, and then everybody copies it.

[mendes-jom]

Thanks a lot for your intervention.
Merci beaucoup
Muito Obrigado

[mendes-jom]

And Cases by Mio better too.

Thanks again

[Aureuum]

Since they don't seem to fix it, you can just replace your data with this:
4/1/20,4/2/20,4/3/20,4/4/20,4/5/20,4/6/20,4/7/20,4/8/20,4/9/20,4/10/20,4/11/20,4/12/20,4/13/20,4/14/20,4/15/20
56989,59105,64338,68605,70478,74390,78167,82048,86334,90676,93790,95403,98076,103573,106206

[boogheta]

it was fixed earlier tonight! Thanks @CSSEGISandData !

[MelbourneDeveloper]

@LoretoSanchez I've logged a similar problem here. In fact, I have a massive log of all the places where the confirmed count actually goes backwards. There are about 100 instances.

@CSSEGISandData

[MelbourneDeveloper]

@Aureuum where do those replacement values come from?

[Aureuum]

@MelbourneDeveloper They come from here:
https://github.com/opencovid19-fr/data/tree/master/ministere-sante
and here:
https://en.wikipedia.org/wiki/2020_coronavirus_pandemic_in_France

[Aureuum]

@boogheta Yes, they fixed several of the values for France, but the 4 last, for 4/12-4/15, are still wrong.

[LoretoSanchez]

Thank you so much for the replies!

[MelbourneDeveloper]

@Aureuum, @LoretoSanchez have a look at this colossal mistake

	French Polynesia	France	2020-03-23 23:19:21	-17.6797	-149.4068	19874	860	2200	16814	French Polynesia, France

Again, I picked this up using my database tool. I can show people how to use it to find these anomalies.

https://github.com/MelbourneDeveloper/COVID-19-DB

On this day it says that French Polynesia had 19874 cases.

[Aureuum]

@MelbourneDeveloper Yes, big error for sure!
The time series says Fr.Pol. has 55 total cases (but haven't checked that against the official).
And thanks for sharing your database tool! :)

[MelbourneDeveloper]

@Aureuum I would love to help automating the process of validating the data

[Aureuum]

@MelbourneDeveloper Thanks! :)

[RouxRC]

@Aureuum They explain why in #2094. Starting from April 12th, they include both confirmed cases and suspected cases from nursing homes, which are not supposed to overlap anymore since France now reports 2 different figures from these.

[MelbourneDeveloper]

@RouxRC shouldn't data be revised from before that time?

If data is not revised, it's impossible to do calculations over time and time-based modeling is next to impossible.

[RouxRC]

I agree, but the data does not exist so far, this is not @CSSEGISandData's responsability until such data is actually released by France for this period (which I fear will probably never come). Although IMHO what we could discuss is the choice of including suspected cases in an indicator named "confirmed", which for most other countries only accounts for actual positive tests.

[MelbourneDeveloper]

@RouxRC the column is already called "Confirmed". What I suggest is a "Suspected" column. That would work well because it would allow any country that is recording suspected cases to add data there.

[Aureuum]

I also think it's important to be able to compare countries on equal terms, so we would at least have a dataset for FR as well with just confirmed cases.

[MelbourneDeveloper]

@Aureuum true, but the point is the column is called Confirmed. Suspected cases have no business being in a column called Confirmed.

[Aureuum]

And thanks @RouxRC for the clarification! :)

[Aureuum]

@MelbourneDeveloper Yes of course, that's important to avoid any confusion.

[MelbourneDeveloper]

@Aureuum @RouxRC here is a comprehensive list of all the places where the count is revised down:

#2256

There are some very big discrepancies and someone should work on fixing this. At the very least, there should be a comment to note why there is a discrepancy.

[Aureuum]

@MelbourneDeveloper Thanks for providing this list! Yes, that would be appropriate with explanations.

[MelbourneDeveloper]

@Aureuum very happy to be of service. Can you point to other kinds of discrepancies that need validation?

[Aureuum]

@MelbourneDeveloper
Thank you.
I'm not aware of any other at this moment.

[cipriancraciun]

I have reported this a few days ago on #2094 (comment)

And although they've closed that issue, the France / French Polynesia numbers are still wrong.