search

COVID19Tracking / website

The COVID Tracking Project website
https://covidtracking.com/
Apache License 2.0
500 stars 190 forks source link

Provide simple visualizations, link to existing ones in the meantime #91

Closed nealmcb closed 4 years ago

nealmcb commented 4 years ago

Thanks for this project!

Adding visualizations and the ability to query and drill down into the data would be very handy. E.g. I'd love to see a time series plot of the daily data for just one state or one outcome (positive or death or whatever).

Until that is available, thanks for linking via Twitter to one visualization website for this data:

That one is great because it shows barcharts of data over time per state. That site is:

I've seen the URL fail a few times with a warning from Tableau, but if it remains stable, you might want to list it from the website.

For comparison, I also note some other visualizations, which as far as I've seen don't provide the ability to drill down to a region and see a time series for just that area:

joshuaellinger commented 4 years ago

Can you replicate this on the website repo and close it here when done?

joshuaellinger commented 4 years ago

moved from the general issues to web site.

on2air commented 4 years ago

I run an agency around Airtable and decided to pull in your data to Airtable which is pretty good data visualization: https://airtable.com/shrGZit91PisYqpdM/tblvj81hpMq0uZ1FB/viw7h7DDzmL0u4q5p?blocks=bipNeGkkQAz6fvcaA&bip=full

Feel free to share. It requires a code snippet on my end to run each day and pull in your updates that I'll try to keep updated.

nealmcb commented 4 years ago

@on2air Cool - thanks! Nice stacked bars etc. But you seem to be showing cumulative sums of values that are already cumulative totals-to-date. E.g. the current total for NY State is 7,102, but you're displaying 18,949.

on2air commented 4 years ago

ah thanks @nealmcb . I've updated to calculate daily numbers. should be reflected now

theojulienne commented 4 years ago

https://covid.graphics/country/usa.html (and sub-state pages) are now pulling the wonderful (:heart: :heart: thank you!) data provided from the API here for US states over the top of JHU CSSE data for the rest of the world.

nealmcb commented 4 years ago

Note the excellent visualization project now at https://github.com/COVID19Tracking/covid-tracking-dash

I'd love to see it provide derivatives: daily increments for each parameter, especially for the hospitalization data, as noted at How to fight the coronavirus SARS-CoV-2 and its disease, COVID-19 Michael Lin, PhD-MD

MrSteve2 commented 4 years ago

@nealmcb are you looking for something like they talk about on slides 14/15 "Estimated 5% of total infected (not diagnosed cases) require hospitalization" Ie Chart positives / hospitalizations? Like this for NY visualization-1

I can quickly create Vega/Vegalite versions that could be auto-generated and could be used to insert into your website. IE, if there is anything I can do to help, let me know.

nealmcb commented 4 years ago

@MrSteve2 Thanks for the offer! I'm looking at page 15:

The next few weeks are critical: We must do whatever we can to reduce R0 or increase doubling time. Social distancing efforts in CA started 3/20. Since cases are undercounted, we will only know if efforts are working by looking at weekly hospitalization trends 2–3 weeks later (4/10).

If hospitalizations are still doubling each week on 4/10, we have only another 3 weeks to get a second chance.

If that doesn’t work by 5/1, there will be no third chance. We would have to immediately clamp down to avoid hospital overflow. This would require Wuhan-like measures such as central quarantine for sick and enforced home-isolation for everyone.

So they aren't tracking the ratio of positives / hospitalizations. They don't trust the data on testing since the criteria for and availability of tests varies over time. They see the hospitalization data as more reliable over time. So just plotting the API column hospitalizedIncrease by day would show if social distancing is flattening that curve. Lots of folks are already plotting some of the other columns by day, and I think plotting all the ...Increase values would be helpful, and especially hospitalizedIncrease.

MrSteve2 commented 4 years ago

@nealmcb I'll see what I can do.
I also found this site: https://covid19.healthdata.org/projections Which has hospital data on (bed and ICU occupancy, and ventilator use).

MrSteve2 commented 4 years ago

OKay I graphed all the data items and added a dropdown to select state. Check https://mrsteve2.github.io/covidTrackingViz.github.io/covidCharts.html

Let me know if the increase graphs were what you were looking for.

nealmcb commented 4 years ago

Thanks, MrSteve2. Those look good! If they were smaller, or combined in cases where the scales are amenible, we could see more of it at once. And I'd put the state selector a the top.

Can the code for that be integrated with the project and dash?

I'll also note a few other links to visualizations that I've been collecting:

MrSteve2 commented 4 years ago

Thanks, MrSteve2. Those look good! If they were smaller, or combined in cases where the scales are amenible, we could see more of it at once. And I'd put the state selector a the top. Agreed. Just wanted to do first step and create. UX design is needed

Can the code for that be integrated with the project and dash?

Yes. Its a simple matter of programming :) Do you have a particular project(s) I can clone and page/pages you would like to see them integrate with? If so let me know and I will take a look at what it will take.

I'll also note a few other links to visualizations that I've been collecting:

* covid.graphics: https://covid.graphics/country/usa.html

* COVID on Worldmapper: https://worldmapper.org/?s=covid

* Coronavirus Deaths by U.S. State and Country Over Time: Daily Tracker - The New York Times: https://www.nytimes.com/interactive/2020/03/21/upshot/coronavirus-deaths-by-country.html

Thanks will check and perhaps do a review of them to determine what we can learn and what we should steal (with credits of course)

phact commented 4 years ago

I made this viz based on the John's Hopkins data, would be happy to contribute something similar if there is interest.

https://covid-19-keppler.netlify.com/