Corona Virus Data Points - gather data and facts and make them visible.

License: MIT License

Python 93.81% HTML 6.19%

covid-19 covid19 covid-virus covid-data covid covid-2019 covid19-data covid-19-italy covid-19-germany covid-19-france

ncov19-datapoints.github.io's Introduction

nCov19-DataPoints

Corona Virus Data Points - gather data and facts and make them visible.

Purpose

I found that most maps visualizing the Corona Virus data while very interesting did not help me judging the situation in local places - the resolution was too rough, and the numbers were in no relation to the population in the affected area.

So I set this map up with this repository.

Feedback and help

If you think something is wrong or missing from the map, please add it to the github issue tracker. There is still plenty of stuff to do.

If you would like your country or region to be added, you can help by finding the required sources of data - preferably csv files by an official source, but other people gathering and cleaning up data with clear source attribution are also welcome. Or, if you feel like it, take the existing Python scripts for countries as an example and create your own one and send me a pull request. All help is welcome!

I will add more detailed help soon, until then, just ask.

Repository structure

The structure of the repository is currently very simple:

index.html: the starting page if you go to http://ncov19-datapoints.github.io/, displaying the map with all the geojson files generating by the Python scripts for each country.
For each country representing a separate data set with separate data sources there is a folder containing the following files (as an example the links go to the UK files):
- build.py: the script run each day at 20:00 UTC to update the ncov19map.js file.
  - There is a helper functions "generateOutput" in buildhelpers.py. Use that to save your files to make sure the map can use them even if there are any style changes etc. I will add a few checks in there later to make sure the data can be used by the map.
- ncov19map.js: the file used by the map in "/Cov19.html"
- ncov19.geojson contains the same data as the ncov19map.js file, just without the minor JavaScript additions which make the data digestible to the Cov19.html file.
- ncov19.csv contains the raw nCov19 data, this is useful for debugging. Or maybe somebody else wants to do something with that data.

Thanks

Many thanks must go to the people gathering all the information which makes projects like this possible:

Of course the JHU repository is awesome.
For France the OpenCOVID19 France repository is a very good data source! They also have a very nice dashboard for their data. I just wish the departements would not lag the regions, but that is complaining on a very high level.
For Germany, Spain, Italy and England there are official data sources in a digestible format - a hurray to the civil servants doing that:
- Italy: http://github.com/pcm-dpc/COVID-19
- England: http://www.arcgis.com/home/item.html?id=e5fd11150d274bebaaf8fe2a7a2bda11 - though it would be nice if England would do the data more fine-grained - areas with a population of more than 1 million tend to obscure hotspots in the map.
- Spain: https://covid19.isciii.es/ - also kind of large areas, obscuring hotspots.
- Germany: https://npgeo-corona-npgeo-de.hub.arcgis.com/datasets/917fc37a709542548cc3be077a786c17_0/data (or http://corona.rki.de)
The Swiss data can be found here: https://github.com/openZH/covid_19 (again thanks!).
- If you need the data, feel free to use the csv-file. I aim to update it once a day, usually in the evening, to the best of my abilites. Once the RKI publishes the information, I will use that as a source, and the CSV file will then just contain the RKI data.
For the US, the NY Times has graciously decided to make their data available to the public:
- US: https://github.com/nytimes/covid-19-data
Many thanks to all the people who refined the geojson data and made it ready-to-use!
And http://github.com/mbloch/mapshaper is a very cool tool, if the too detailed geojson brings the browser to a crawl.
Many thanks to http://openstreetmap.org - without them, no map!
And also to the Leaflet project, making it really easy to display geojson files!
Also of course many thanks to github - this is an awesome platform with lots of free stuff!

ncov19-datapoints.github.io's People

Contributors

Stargazers

Watchers

ncov19-datapoints.github.io's Issues

Make map hierarchical

Add complete world map using JHU data.
Use more detailed data where available.
Use topojson insteead of geojson for the necessary map
What to do if the detailed data is only available for some parts and not at all for other?
What to do if the data for the parts is older/newer than the data for the whole?
What to do if the data for the parts is inconsistent with the data for the whole?
First solution: ignore any inconsistencies, display all detailed data which is available on top of the summary data (or better: use the summary data to fill out all the missing parts of the more detailed data).

Add a timeline

That would be nice!

Improve buildhelpers.generateOutput

The function needs plausibility checks etc to make it possible to figure out if a build.py script does the right thing.

Obviously there is also a lot of room to improve the general structure of the Python scripts (i. e. buildhelpers could import the individual build.py scripts and execute a set of functions).

Fix the "source" link in the info pane

While the "source" link itself works, it is impossible to click on it because the info pane is changed whenever the mouse cursor changes position. Doh!

Make info pane and legend vanish in mobile view

Mobile view works, but the info pane and the legend cover half of the screen! They need to be collapsed or something...

Use chromjs for color scales?

https://github.com/gka/chroma.js

Switzerland

Interesting because it looks like the infection went through it from Northern Italy to south-east France.

Also might be another example like South Korea of a country getting the stuff under control.

Data can be scraped here: https://www.bag.admin.ch/bag/de/home/krankheiten/ausbrueche-epidemien-pandemien/aktuelle-ausbrueche-epidemien/novel-cov/situation-schweiz-und-international.html

Add China (or at least Hubei) and/or South Korea

Those will be really interesting to watch. What happens if live if returned to normal after a shutdown?

Finish Austria

Austria has been started, but the data source is a HTML page and scraping is work!

USA

The US will most certainly be one of the most interesting cases to watch and learn from.

Interesting link which might yield the necessary data:
https://github.com/lazd/coronadatascraper

Note: it might be wise to do some sort of fallback solution depending how much data is available: state-based data (available from the JHU github page) if nothing with smaller granularity is available.

Automate German data gathering

Either scrape or hopefully the RKI will publish a machine readable list on Landkreis level some time...

Fill outdated regions with a pattern

To indicate stale data, put a dashed pattern over regions older than 24 hours

Visualize how much confirmed cases might reflect actual cases based on reported deaths

For a lot of countries we can assume that the number of deaths by infection with nCov-19 is a fairly accurate reflection of the actual situation.

However we cannot assume the same for the number of reported confirmed cases - coverage here clearly wildly differs between different countries, e. g. compare South Korea and Italy.

The number of reported deaths however gives a very rough idea about the number of actual cases around 12 days ago (the rough average time between infection and death, if death occurs, need to find the paper with that number again). Assuming a mortality rate of about 2% (as long as the health system is not overwhelmed that seems still high), we can make a back-of-the-envelope estimate, that 12 days ago, the number of actual infected people (+recovered people) was about 50 times the number of dead people until today.

If you take that simple measuring stick to Italy (or actually quite a lot of other countries), it is obvious that only a fraction of the actual cases have been diagnosed and reported as confirmed cases.

It would be nice to translate into some sort of confidence measure displayed on the map. Probably better not to use numbers, because this is all just an educated guess, but as a plausibility check it can inform people if the map reflects the actual situation on the ground.

ncov19-datapoints / ncov19-datapoints.github.io Goto Github PK