search

aatishb / covidtrends

Tracking the growth of COVID-19 Cases worldwide
https://aatishb.com/covidtrends/
MIT License
301 stars 107 forks source link

What's going on with Ecuador? #185

Open git2samus opened 4 years ago

git2samus commented 4 years ago

The confirmed cases seem to be decreasing and the chart went out of bounds.

https://aatishb.com/covidtrends/?trendline=false&location=Ecuador

MrSpiffyClean commented 4 years ago

I'm guessing there was a recheck of cases in Ecuador and some cases that were overcounted before are now being rectified, which is why the graph seems to "go backwards" around 2020-05-05. Need to check the actual data (and any news reports perhaps) but it seems to be the most logic explanation. Console view of plot data of Ecuador

MrSpiffyClean commented 4 years ago

Looking about a bit I found some reports confirming what I stated. From https://www.eluniverso.com/noticias/2020/05/08/nota/7835562/casos-coronavirus-ecuador-viernes-8-mayo-11h00-30-confirmados-16:

La cifra de contagiados volvió a bajar en -1480 casos-, Zevallos volvió a decir que el sistema de procesamiento de datos sigue ajustando duplicidad y triplicidad de resultados en una misma persona, pues el sistema que se utilizaba tiene diez años de antigüedad.

Translating:

The [Confirmed Cases] count went back again, to -1480 cases, Zevallos again said that the data processing system keeps counting double and triple results to a single person, since the used system is ten years old.

As for why the graph looks that way, it all has to do with how the data is displayed. The Y axis represents the difference between the number of confirmed cases on the selected date and the number of confirmed cases 7 days before. If you compare the number of cases in 2020-05-04 and 2020-05-11, you'll notice that there are less confirmed cases now than before, hence the drop.

SteveCadaver commented 2 years ago

Hi Gents. Seems to be a problem not related to the rolling week count.

As shown here Total Reported Deaths for South Africa on 2021-11-14: 89746 and the next day 2021-11-15: 89489. Where do you pull the data from? image

I've been following the stats here, and its a great tool for keeping track of Covid-19.

MrSpiffyClean commented 2 years ago

The whole "graph jumping up and down thing" is, as I mentioned earlier, a consequence of values going down in what should be an always growing series of values. And this can be usually attributed to post-fact corrections to the dataset, to which I started checking some news sources and the like... however, for the case of South Africa, no such thing appeared in that specific timeframe. (there is a recent report about an antigen test backlog that was added, but that is too recent (check https://coronavirus.jhu.edu/region-data-notes/south%20africa)

Thus, I guess the most likely explanation is that someone messed up entering the values for the South Africa dataset. Checking the official figures (at https://sacoronavirus.co.za/category/daily-cases/) does show values that do make some sense... The data for the graph comes from the John Hopkins Covid-19 dataset (https://github.com/CSSEGISandData/COVID-19) (apart from the US, which might use the NY Times dataset). I'll report the issue to them as soon as I'm able, though there's no guarantee that it will get fixed.

MrSpiffyClean commented 2 years ago

After a bit of time, the folks at the John Hopkins Covid-19 dataset read and dealt with the issue I raised, and it seems to have fixed the issue.

covidtrends_sa_fix