Have street address, need geocoordinates using Python and geopy

user_7124155 0
Viewed 243

I have a data set of street addresses and need to determine if they are within certain lat/long coordinates. Since I'm using Python, geopy seems like a common way to go.

But geopy is not a geolocation service, it's "just a library which provides these implementations for many different services". So I see two options:

  1. Use geopy with Nominatim, which is free but you get only one request per second (it's slow)

    from geopy.geocoders import Nominatim

or

  1. Use geopy (or not) with a paid service that is faster, such as Google, PickPoint, or something else.

Do I have the correct understanding? (are there any fast free options?)

geopy geolocation python-3.x
1 Answers

There are a lot of ways to accomplish this.

The most common is to use GoogleCloud service (it is indeed free but demands a bank card) & the package googlemaps

import googlemaps

gmaps = googlemaps.Client(key='Add Your Key here')

# Geocoding an address
geocode_result = gmaps.geocode('1600 Amphitheatre Parkway, Mountain View, CA')

#Here you can process your result at your leisure
lat, lon = geocode_result.geometry.location.values()

But there is a lot of projects without necessity to give your bank personal information, such as geocoder. It can use the whole list of geocoding services. For example, ArcGis is free & fast enough (not a request per second)

import geocoder

result = geocoder.arcgis(location="Regency Run, San Antonio, Texas")
if result.ok:
    lat, lon = result.latlng
else:
    raise Exception("I have gotten a bad result :-(")

The presence of point inside perimeter you can define, for instance, with geopandas

import geopandas as gpd
from shapely import Point

gpd.io.file.fiona.drvsupport.supported_drivers['KML'] = 'rw'

some_map = gpd.read_file(fr'some_map.kml', driver='KML')
some_map.contains(Point(lon, lat)).values

P. S. There is a slight trick you could use. Parallel your requests & use several services simultaneously (ArcGis, Google, Yandex &c.). The results are not consistent strictly but maybe it is not a problem for your task with several millions addresses.

user_173091080
Related