Integrate mapclassify

[kylebarron]

From the geopandas.plot docstring

Name of a choropleth classification scheme (requires mapclassify). A mapclassify.MapClassifier object will be used under the hood. Supported are all schemes provided by mapclassify (e.g. ‘BoxPlot’, ‘EqualInterval’, ‘FisherJenks’, ‘FisherJenksSampled’, ‘HeadTailBreaks’, ‘JenksCaspall’, ‘JenksCaspallForced’, ‘JenksCaspallSampled’, ‘MaxP’, ‘MaximumBreaks’, ‘NaturalBreaks’, ‘Quantiles’, ‘Percentiles’, ‘StdMean’, ‘UserDefined’). Arguments can be passed in classification_kwds.

[davidbailey]

The following worked for me:

import numpy

from lonboard import Map, SolidPolygonLayer
from mapclassify import JenksCaspall

layer = SolidPolygonLayer.from_geopandas(zones_interpolated, opacity=0.2)
jc = JenksCaspall(
    zones_interpolated["ProductionAttractionSum Density"].fillna(0),
    k=20
)
colors = []
for color in jc.yb:
    colors.append(Viridis_20.colors[color])
layer.get_fill_color = numpy.uint8(colors)

Map(layers=[layer])

[kylebarron]

Thanks! By the way, I'm not sure where Viridis_20 comes from, but doing an array lookup would be much faster than a loop. You should probably be able to do something like

Viridis_20.colors[jc.yb]

[knaaptime]

since this is still open, i've cooked up a proof of concept that uses pandas's extension API to create a geopandas explore method that uses lonboard as a backend. The code is here and an example notebook here. The functions are fairly small and dont do much besides handle classification schemes and categorical variables using mapclassify/matplotlib and expose a few more styling args for convenience and parity with the existing explore

works well for my use cases and I'm happy to keep it in geosnap if it's out of scope. That said, I suspect it might be more useful/visible in a central location like lonboard itself so also more than happy to contribute it if anyone has interest

[knaaptime]

heres a little notebook that updates the color scheme dynamically using jupyterwidgets (note, not all classifiers take advantage of all k)

https://gist.github.com/knaaptime/cc6863d99a4a08a9825191dc27b3a5f1

[kylebarron]

Thanks @knaaptime, that's very cool.

I don't think I want to add mapclassify as a required dependency of Lonboard, but I'm open to having a GeoPandas integration like it's mentioned that hvplot does. So then we could have

from lonboard.geopandas import register_plot_methods

register_plot_methods()

Then

gdf = gpd.GeoDataFrame(..)
gdf.plot()
# or maybe
gdf.lb.explore()
# or even have a way to override .explore

Would you be open to making the PR?

Then we could have the mapclassify code be self-contained within the GeoPandas integration.

[knaaptime]

cool, that's exactly what i was thinking (agree, no desire to make mapclassify a hard dependency). Happy to put together the PR

[knaaptime]

i'll start the PR so the code is more visible, but for some reason I cant import the local version of lonboard because it cant find /lonboard/lonboard/static/index.css and .js

[knaaptime]

(nvm, i just grabbed those files from the tarball on pypi; looks like maybe they never got committed or something)

[kylebarron]

https://github.com/developmentseed/lonboard/blob/main/DEVELOP.md