search

dsfsi / covid19za

Coronavirus COVID-19 (2019-nCoV) Data Repository and Dashboard for South Africa
https://dsfsi.github.io/covid19za-dash/
MIT License
255 stars 201 forks source link

[DATA] Blip in the data #920

Open Diederikjh opened 2 years ago

Diederikjh commented 2 years ago

It looks like there is a bit of a blip in the test data.

See attached screenshot

Screenshot_20211124-063022

vukosim commented 2 years ago

Thanks @Diederikjh will check

lostpebble commented 2 years ago

This is a pretty important fix that needs to be done. It seems Google is using the data from this repo- and with this new variant and all the news around it, when people search for "South Africa Covid" this graph with the blip shows up now. It looks very scary, as if this new variant has created a sudden crazy surge, and its likely having real repercussions.

krokkie commented 2 years ago

It appears as if there were some historic revisions to the data, artificially raising the number of positive cases by 18,000 from 23-Nov onwards. The number here: https://sacoronavirus.co.za/2021/11/23/update-on-covid-19-tuesday-23-november-2021/ reports 2.948m cumulative cases, where as the previous day it was 2.930m. See: https://sacoronavirus.co.za/2021/11/22/update-on-covid-19-monday-22-november-2021/
However, the "new cases" number does not reflect the 18k increase, it shows less than a 1000.... for the 23rd.

thomaslane commented 2 years ago

Have been using this data for my website, what is the best way to handle revisions such as this for the graphical display of accurate daily new cases? Would it not be beneficial to add to the data set the officially reported daily new confirmed cases in addition to the cumulative total confirmed cases?

vukosim commented 2 years ago

@thomaslane Thanks so much for your concern. Its just that time of the year in teaching and exams so I have not had the time to give full attention to make a suggestion. Solid suggestions are welcome and we can involve some of the very active contributors such as @krokkie and @lrossouw

Diederikjh commented 2 years ago

Another idea: I've seen Time magazine's covid newsletter including backdated adjustments in the total infection number, but not the daily infection rate number. They did have an asterisk next to that row in the table for a few weeks,to explain the anomaly.