Open cpyic opened 4 years ago
@cpyic If you want U.S. county data consistently identified by FIPS, consider using data from USAFacts.org. They get data directly from county governments. See the data for Nantucket and Dukes in this screenshot.
Aside from Kansas City, as you pointed out, there's also at least one other difference and that's for NYC. The five boroughs of NYC from the JHU database having zeros in four of the five FIPs. NY County (Manhattan) has the sum of all five boroughs for reasons that I don't quite understand.
You can mouse over every county here to see the latest numbers (total and per-capita) using data from USAFacts.org. Type the name of the county you're interested it in the search box and the display will zoom into that county.
I uploaded the code to use both USAFacts.org and JHU CSSE data. If you really want to dig into the differences between USAFacts.org and JHU CSSE (at least for U.S. cases), you can run the code yourself or scroll to the bottom of these Jupyter notebooks with nbviewer: https://github.com/jjbenes/covid19. You'll be able to fly around the map to see the data.
Hope this bit of info helps.
Want to ask if the four entry without FIPS code were exclusive from their representative FIPS counterparts.
I have run into the same issue when trying to build county-level maps with Qlik and Tableau BI tools. These tools simply do not like having cities and counties in the same map, especially when a city sprawls across multiple counties. I cannot get Kansas City, Missouri to render at all.
If CSSE would confirm that the Kansas City numbers are already contained in its four counties, then I can discard Kansas City as an entity and just plot its counties instead. The virus doesn't pay attention to the lines on the map anyway!
If CSSE would confirm that the Kansas City numbers are already contained in its four counties, then I can discard Kansas City as an entity and just plot its counties instead. The virus doesn't pay attention to the lines on the map anyway!
So I did the math:
The four counties of Jackson, Clay, Cass and Platte have a total of 710 cases as of May 19.
The entity "Kansas City, Missouri" has a total of 902 cases despite being smaller than the four counties together!
So no idea which one is correct. I am inclined to filter out "Kansas City, Missouri" since it is impossible to map it using much commercial software.
@kevinp2 Do you have to use JHU data? I didn't want to have to filter data, so I used USAFacts.org. I got 50 states and DC. Then I added some counties together to get metro areas. See this page: http://first-principles.ai:5100/compare_states.
Want to ask if the four entry without FIPS code were exclusive from their representative FIPS counterparts.
For "time_series_covid19_confirmed_US.csv" there were four entry without fips, including:
For 1., numbers for the separate Dukes and Nantucket counties were 0 in the 4/18 update, but "Dukes and Nantucket" reported 23. This seemed fine as they were mutually exclusive and would not be counted twice when summed.
For 2., Based upon wikipedia Kansas City included the four counties listed above. But each individual 4 counties all reported numbers (268, 53, 27, 56), which when added were very close to 412 of Kansas City on 4/18. Then Kansas City was counted twice when summed up for Missouri. Or were these mutually exclusive?
Would 3. and 4. be the same as for 1., so the FCI and MDOC were not added to the county where they were located?
Many thanks!