I’m jumping into maps a bit and seeing that a lot of the templates require FIPS codes to, for example, make a county-level choropleth. I’m wondering how I could go from long/lat coordinates to FIPS codes in JS. I’m coming from R and there’s an example I found that can take coordinates and convert to GEOID’s (fips). Is there a way to do this kind of thing all in Observable?
General Background: new to maps and thought the choropleth was an easier starting point than projections. Got other pointers? School me, please!
I’ve made quite a few choropleths and never really had the need to send lon/lat coords to FIPS code. The whole point behind FIPS codes is to ensure that you’ve got some standard way to associate data with geographic entities.
Here’s a pretty simple choropleth that I built recently as an ad for a class I’ll be teaching next semester. There are two FileAttachments:
contiguous_states_tigerline.json, which is a topojson describing the geography and
pop.csv, which is tabular data with populations.
Both files refer to FIPS code so it’s easy to attach the data.
Very nice! I guess I’m curious about using some process when I have data as coordinates rather than FIPS. Do you have a go-to example/approach for plotting coordinate data?
Sure - You’ve probably got a projection for the map so you just apply that to the points in your data. Here’s an example where I assume that you want to plot a list of cities as points on a map of the US and the coordinates of those points are stored as lon/lat data. The process would be similar, though, if you want to any other type of object.
The R example to which you linked seems to use a kind of reverse geocoding, where you have a point and you want to find the shape (.e.g., county, state, tract) in which that point falls. This process can end up with a choropleth, though it’s more of a data management step than a visualization one.
In my work, I often have a bunch points which represent the locations of healthcare providers, so I reverse them to individual counties or census tracts using a shapefile. At this point, I have a FIPS code as a result of the reversing process. I aggregate providers with the same FIPS code and divide them by the general population of the containing shape/geography to derive a rate. I then make a choropleth of the rate.
To @mcmcclur’s point, the key thing is that I need to end up with a dataset that has FIPS codes (at least with US data) in order to map those rates to colors and geographies.
The nuts and bolts of this can be done several different ways, but one of the easiest javascript tools I’ve found for the “reversing” part is which-polygon, one of Volodymyr Agafonkin’s many excellent contributions to js mapping. I’ve used the library here, for instance.