location intelligence focuses on easy and efficient map, routing and GPS/trace data collection, processing, analysis and learning.
Unambiguous abstractions for location intelligence entities + converters.
Avoid implementing boilerplate geospatial classes such as GeoPoint, Point or Route (and conversions to/from popular formats like GeoJson) across your organziation's codebase.
Furthermore establish and agree on fixed semantics for each entity in the domain, for free-flowing and unambiguous discoure and discussions with peers.
Efficient procedures for projection-aware processing and transformation of location entities (e.g. routes) - no need to stack pyproj over shapely or to depend on the heavy machinery introduced by geopanmapbox.
Example: add noise to Geometry
from locintel.core.datamodel.geo import Geometry
import matplotlib.pyplot as plt
import numpy as np
geo = Geometry.from_geojson('your_geometry.json')
geo_gps = geo.add_noise()
geo_very_noisy = geo.add_noise(func=np.random.normal, loc=1, scale=50)
plt.scatter(*zip(*geo.to_lng_lat_tuples()), s=1, c='silver')
plt.scatter(*zip(*geo_gps.to_lng_lat_tuples()), s=1, c='g')
plt.scatter(*zip(*geo_very_noisy.to_lng_lat_tuples()), s=1, c='r')
Easy-to-use interfaces to harvest data from variety of services (e.g. google/bing routing, matching, geocoding, etc.) - be free from crunching and managing API semantics into urlllib or requests calls.
Example: harvest route
from locintel.core.datamodel.routing import RoutePlan, WayPoint
from locintel.harvest.routes import calculate
rp = RoutePlan(Waypoint(20.0, 10.0), WayPoint(15.1, 10.1), mode='car')
mapbox = calculate(rp, 'mapbox')
google = calculate('google')
# Access important route information through common interface
mapbox.distance
mapbox.duration
google.distance
google.duration
# Convert to other popular formats
mapbox.geometry.to_geojson()
mapbox.geometry.to_poyline()Example: harvest route (more control)
from locintel.core.datamodel.routing import RoutePlan, WayPoint
from locintel.harvest.routes import MapboxRouter, GoogleRouter
rp = RoutePlan(Waypoint(20.0, 10.0), WayPoint(15.1, 10.1), mode='car')
mapbox_route = MapboxRouter(username='username', password='password').calculate(rp)
google_route = GoogleRouter().calculate(rp)
# Access important route information through common interface
mapbox_route.distance
google_route.distance
# Convert to other popular formats
mapbox_route.geometry.to_geojson()
mapbox_route.geometry.to_poyline()Suite of tools for processing and analysis of routes and respective travel times.
Example: generate random route plan in polygon
from locintel.routes.generators import RandomRouteGenerator
import shapely.geometry as sg
berlin = sg.Polygon([(13.281949, 52.542348), (13.509650, 52.542348),
(13.509650, 52.482488), (13.281949, 52.482488)])
rp = RandomRouteGenerator().generate_route(polygon=berlin)Example: compare two geometries
from locintel.core.datamodel.geo import Geometry
from locintel.routes.metrics.geometry import GeometryComparator
geo_1 = Geometry.from_geojson('geo1.json')
geo_2 = Geometry.from_geojson('geo2.json')
score = GeometryComparator.compare(geo1, geo2, method='hausdorff')Example: analyse geometries
from locintel.core.datamodel.geo import Geometry
geo = Geometry.from_geojson('geo.json')
geo.has_loops()
geo.skewness() # Straight-line-distance/interpolated lengthExample: benchmark routes
from locintel.core.datamodel.routing import RoutePlan, WayPoint
from locintel.routes.metrics.geometry import hausdorff_distance
from locintel.harvest.routes import calculate_competitive
rp = RoutePlan(Waypoint(20.0, 10.0), WayPoint(15.1, 10.1), mode='car')
results = calculate_competitive(rp, ['mapbox', 'google'], comparators=[hausdorff_distance])
# save test results into CSV
results.to_csv('results.csv')Tools supporting the processing and handling of trace (GPS) data (e.g. loading, filtering, sampling).
Example: filter out low confidence GPS points
from locintel.traces.filters import LowConfidenceFilter
traces = DruidLoader(**config).load()
filter_policy = LowConfidenceFilter(min_confidence=0.5).filter
filtered_traces = filter(filter_policy, traces)Example: re-sampling
from locintel.traces.samplers import SimpleSampler
trace = next(DruidLoader(**config).load())
resampled_trace = SimpleSampler(period=5).sample(traceGraph abstraction, processing and manipulation.
With the wealth of geospatial library out there, it is important to underline what this library is not intended to be: