NYC With GeoPandas

Source

This is a look at getting a map of the New York City boroughs done using GeoPandas. The GeoPandas examples come from the Introduction To GeoPandas.

Setup

Imports

Note: Using hvplot requires geoviews and scipy on top of geopandas.

# python
from functools import partial 

import sys

# pypi
from tabulate import tabulate

import altair
import geopandas
import hvplot.pandas

# my stuff
from graeae import EmbedHoloviews
from graeae.visualization.altair_helpers import output_path, save_chart

TABLE = partial(tabulate, tablefmt="orgtbl", showindex=False)

SLUG = "nyc-with-geopandas"
OUTPUT_FOLDER = f"files/posts/{SLUG}"
Embed = partial(EmbedHoloviews, folder_path=OUTPUT_FOLDER)

OUTPUT_PATH = output_path(SLUG)
save_altair = partial(save_chart, output_path=OUTPUT_PATH, height=650)

Versions

print(sys.version)
print(altair.__version__)

3.9.9 (main, Dec 21 2021, 10:03:34) 
[GCC 10.2.1 20210110]
5.0.0rc1

I'm using the pre-release version of altair since its interface is different from the 4.x versions.

The Data

GeoPandas comes with some built-in datasets (only three, actually) including one for the Boroughs of New York City, which is what we'll load here.

Here's the available datasets.

for dataset in geopandas.datasets.available:
    print(f" - ~{dataset}~")
  • naturalearth_cities
  • naturalearth_lowres
  • nybb

We want the last one, the nybb. To load the map data into GeoPandas we give it the path to the source file. In this case, since it comes with the built-in data sets we can use their get_path method.

path_to_data = geopandas.datasets.get_path("nybb")
nyc_data = geopandas.read_file(path_to_data)

print(nyc_data.head(1))

   BoroCode       BoroName     Shape_Leng    Shape_Area  \
0         5  Staten Island  330470.010332  1.623820e+09   

                                            geometry  
0  MULTIPOLYGON (((970217.022 145643.332, 970227....

The GeoPandas methods will automatically use the geometry column when we do any mapping so we don't have to do anything special to tell it what to use. Note that although printing the first row makes it look like we have a pandas DataFrame, it's actually a GeoPandas object, so it has both pandas methods and geopandas methods.

Plotting The Boroughs With HVPlot

plot = nyc_data.hvplot(hover_cols=["BoroName"], legend=False, tools=["hover", "wheel_zoom"],).opts(
    title="New York City Boroughs",
    width=800,
    height=700,
    fontscale=2,
    xaxis=None,
    yaxis=None
)
outcome = Embed(plot=plot, file_name="nyc_boroughs")()

print(outcome)

Figure Missing

Folium Map

The GeoPandas explore method creates a Folium map that will plot the boroughs using the geometry data and put it on top of a street map. To give it something extra we'll add a column for the area of each burough and color the output based on it.

nyc_data["Area"] = nyc_data.area
print(nyc_data.Area)

0    1.623822e+09
1    3.045214e+09
2    1.937478e+09
3    6.364712e+08
4    1.186926e+09
Name: Area, dtype: float64

explorer = nyc_data.explore(column="Area", legend=False,
                            tooltip="BoroName", popup=True)
explorer.save(f"{OUTPUT_FOLDER}/nyc-explore.html")

Figure Missing
  • The column tells geopandas which column to use to color the Choropleth Map
  • popup makes it so that clicking on a borough will show data from all the columns

An Altair Map

Using altair isn't too far from using hvplot, although it does have (a little) more documentation for mapping. For some reason it doesn't recognize our data as geographic data so we'll have to use the project method to tell altair to treat the geometry data as x-y data.

chart = altair.Chart(nyc_data).mark_geoshape().encode(
    color="Area",
    tooltip=[altair.Tooltip("BoroName", type="nominal"),
             altair.Tooltip("Area", type="quantitative", format=".2e")]
).project(type="identity", reflectY=True).properties(
    width=800,
    height=525,
    title="NYC Borough Areas",
)

save_altair(chart, "nyc-borough-areas-altair")

Figure Missing

Sources