I'm having difficulty loading the following JSON containing GIS data (https://data.cityofnewyork.us/resource/5rqd-h5ci.json) into a GeoDataFrame.
The following code fails when I try to set the geometry.
import requests
import geopandas as gpd
data = requests.get("https://data.cityofnewyork.us/resource/5rqd-h5ci.json")
gdf = gpd.GeoDataFrame(data.json())
gdf = gdf.set_geometry('the_geom')
gdf.head()
Setting the geometry fails because the geopandas.GeoDataFrame constructor doesn't appear to be built to handle JSON objects as python data structures. It therefore complains about the argument not being a valid geometry object. You have to parse it into something that geopandas.GeoDataFrame can understand, like a shapely.geometry.shape. Here's what ran without error on my side, Python 3.5.4:
#!/usr/bin/env python3
import requests
import geopandas as gpd
from shapely.geometry import shape
r = requests.get("https://data.cityofnewyork.us/resource/5rqd-h5ci.json")
r.raise_for_status()
data = r.json()
for d in data:
d['the_geom'] = shape(d['the_geom'])
gdf = gpd.GeoDataFrame(data).set_geometry('the_geom')
gdf.head()
A disclaimer: I know absolutely nothing about Geo anything. I didn't even know these libraries and this kind of data existed until I installed geopandas to tackle this bounty and read a little bit of online documentation.
For people who are using web mapping libraries...
If the GeoJSON is wrapped in a FeatureCollection, as they often are when exported to a GeoJSON string by web mapping libraries (in my case, Leaflet), then all you need to do is pass the list at features to from_features() like so:
import geopandas as gpd
study_area = json.loads("""
{"type": "FeatureCollection", "features": [{"type": "Feature", "properties": {}, "geometry": {"type": "Polygon", "coordinates": [[[36.394272, -18.626726], [36.394272, -18.558391], [36.489716, -18.558391], [36.489716, -18.626726], [36.394272, -18.626726]]]}}]}
""")
gdf = gpd.GeoDataFrame.from_features(study_area["features"])
print(gdf.head())
Output:
geometry
0 POLYGON ((36.394272 -18.626726, 36.394272 -18....
Easy peasy.
A more idiomatic way that uses regular dataframe functions inherited from pandas, and the native GeoDataFrame.from_features:
gdf = gpd.GeoDataFrame(data.json())
# features column does not need to be stored, this is just for illustration
gdf['features'] = gdf['the_geom'].apply(lambda x: {'geometry': x, 'properties': {}})
gdf2 = gpd.GeoDataFrame.from_features(gdf['features'])
gdf = gdf.set_geometry(gdf2.geometry)
gdf.head()
Combining the above answers, this worked for me.
import pandas as pd
import geopandas as gpd
from shapely.geometry import shape
nta = pd.read_json( r'https://data.cityofnewyork.us/resource/93vf-i5bz.json' )
nta['the_geom'] = nta['the_geom'].apply(shape)
nta_geo = gpd.GeoDataFrame(nta).set_geometry('geometry')
