GeoPandas Support (GeoPandas and Shapely)

GeoPandas GeoDataFrame is a tabular data structure that contains a set of shapes (geometry) per each observation.

GeoDataFrame extends pandas DataFrame and as such, aside from the geometry, can contain other data.

GeoPandas supports the following three basic classes of geometric objects (shapes):

• Points / Multi-Points

• Lines / Multi-Lines

• Polygons / Multi-Polygons

All GeoPandas shapes are “understood” by Lets-Plot and can be plotted using various geometry layers, depending on the type of the shape.

Use:

• geom_point(), geom_pie(), geom_text() with Points / Multi-Points

• geom_path() with Lines / Multi-Lines

• geom_polygon(), geom_map() with Polygons / Multi-Polygons

• geom_rect() when used with Polygon shapes will display corresponding bounding boxes

Plot Preliminaries

import pandas as pd

from lets_plot.geo_data import *
from lets_plot import *
LetsPlot.setup_html()

df = pd.DataFrame({
    'state': ['IL', 'IN', 'MI', 'OH', 'WI'],
    'pop_2021': [12_569_321, 6_805_663, 9_992_427, 11_714_618, 5_852_490],
})
gdf = geocode_states(names=df.state).scope('US').get_boundaries()[['state', 'geometry']]

The geodata is provided by © OpenStreetMap contributors and is made available here under the Open Database License (ODbL).

Use Cases

Depending on the situation, for spatial data we use either parameter data or map or both.

Use either data or map Parameter

Suppose we have the following data:

gdf

	state	geometry
0	IL	MULTIPOLYGON (((-89.20898 37.02010, -89.12109 ...
1	IN	MULTIPOLYGON (((-87.62695 37.92687, -87.62695 ...
2	MI	MULTIPOLYGON (((-88.41797 46.01222, -88.50586 ...
3	OH	MULTIPOLYGON (((-82.17773 38.75408, -82.17773 ...
4	WI	MULTIPOLYGON (((-92.81250 45.02695, -92.81250 ...

If you want to draw only shapes, then it makes no difference which parameter is used:

ggplot() + geom_map(data=gdf)

ggplot() + geom_map(map=gdf)

Use data Parameter

If you want to use aesthetics, the data parameter is the only choice:

ggplot() + geom_map(aes(fill='state'), color='white', data=gdf)

Use Both: data and map Parameters

Suppose you also have a dataframe with population data:

df

	state	pop_2021
0	IL	12569321
1	IN	6805663
2	MI	9992427
3	OH	11714618
4	WI	5852490

In this situation, in order to link aesthetics to the population numbers you will use:

• data parameter for the “population” dataframe,

• map parameter for the state boundaries GeoDataframe.

The 3rd parameter, map_join, will help to combine population values and state boundaries on the same chart:

ggplot() + geom_map(aes(fill='pop_2021'), color='white', data=df, map=gdf, map_join='state')

use_crs Parameter

Specify EPSG code of coordinate reference system (CRS). All coordinates in GeoDataFrame will be projected to this CRS.

ggplot() + geom_map(map=gdf, use_crs="EPSG:32616")

Value “provided” tells Lets-Plot that the input GeoDataframes already contains coordinates in the desired CRS and should not be reprojected any further.

gdf_utm = gdf.to_crs("EPSG:32616")
ggplot() + geom_map(map=gdf_utm, use_crs="provided")

Interactive Maps

Apart from the use_crs parameter, GeoDataFrames are fully supported on Lets-Plot interactive maps.

When used in combination with geom_livemap(), the value of the use_crs parameter is ignored and all coordinates are reprojected using the Mercator projection.

ggplot() + geom_livemap() + geom_map(aes(fill='state'), data=gdf, alpha=.5)

Examples

• The world map with Lets-Plot and GeoPandas:

• Plotting Airbnb prices Boston:

• Spatial prediction of soil pollutants with multi-output Gaussian processes:

• Using geom_imshow() to draw DEM on map:

• use_crs parameter:

• Using exotic map projections:

• An inset map of Kotlin island: