Distance Calculation API¶

Distance matrix computation and geographic distance calculations.

Main Functions¶

allocator.get_distance_matrix(points_from: ndarray, points_to: ndarray | None = None, method: str = 'euclidean', **kwargs) → ndarray[source]¶

Single entry point for all distance calculations.

Parameters:

points_from – Source points (numpy array with shape [n, 2] where columns are [lon, lat])
points_to – Destination points (optional, defaults to points_from)
method – Distance calculation method (‘euclidean’, ‘haversine’, ‘osrm’, ‘google’)
**kwargs – Method-specific arguments: - api_key: Required for ‘google’ method - osrm_base_url: Custom OSRM server URL for ‘osrm’ method - osrm_max_table_size: Chunk size for OSRM requests (default: 100) - duration: For ‘google’ method, return duration instead of distance (default: True)

Returns:

Distance matrix as numpy array with shape [len(points_from), len(points_to)]

Raises:

ValueError – For invalid method or missing required parameters

Distance Modules¶

Factory Module¶

Factory module for distance calculations - main entry point.

allocator.distances.factory.get_distance_matrix(points_from: ndarray, points_to: ndarray | None = None, method: str = 'euclidean', **kwargs) → ndarray[source]¶

Single entry point for all distance calculations.

Parameters:

points_from – Source points (numpy array with shape [n, 2] where columns are [lon, lat])
points_to – Destination points (optional, defaults to points_from)
method – Distance calculation method (‘euclidean’, ‘haversine’, ‘osrm’, ‘google’)
**kwargs – Method-specific arguments: - api_key: Required for ‘google’ method - osrm_base_url: Custom OSRM server URL for ‘osrm’ method - osrm_max_table_size: Chunk size for OSRM requests (default: 100) - duration: For ‘google’ method, return duration instead of distance (default: True)

Returns:

Distance matrix as numpy array with shape [len(points_from), len(points_to)]

Raises:

ValueError – For invalid method or missing required parameters

Euclidean Distance¶

Euclidean distance calculations for geographic coordinates.

allocator.distances.euclidean.euclidean_distance_matrix(points_from: ndarray, points_to: ndarray | None = None) → ndarray[source]¶

Calculate euclidean distance matrix between two sets of lat/lon points.

Parameters:

points_from – Source points (numpy array with shape [n, 2] where columns are [lon, lat])
points_to – Destination points (optional, defaults to points_from)

Returns:

Distance matrix as numpy array with shape [len(points_from), len(points_to)]

allocator.distances.euclidean.latlon2xy(lat: float, lon: float) → list[float][source]¶

Transform lat/lon to UTM coordinate

Parameters:

lat (float) – WGS latitude
lon (float) – WGS longitude

Returns:

UTM x, y coordinate

Return type:

[x, y]

allocator.distances.euclidean.pairwise_distances(X: ndarray, Y: ndarray | None = None) → ndarray[source]¶: Pairwise euclidean distance calculation.

allocator.distances.euclidean.xy2latlog(x: float, y: float, zone_number: int, zone_letter: str | None = None) → tuple[float, float][source]¶

Transform x, y coordinate to lat/lon coordinate

Parameters:

x (float) – UTM x coordinate
y (float) – UTM y coordinate
zone_number (int) – UTM zone number
zone_letter (str, optional) – UTM zone letter. Defaults to None.

Returns:

WGS latitude, longitude

Return type:

(lat, lon)

Haversine Distance¶

Haversine distance calculations for geographic coordinates.

allocator.distances.haversine.haversine_distance_matrix(points_from: ndarray, points_to: ndarray | None = None) → ndarray[source]¶

Calculate haversine distance matrix between two sets of lat/lon points.

Parameters:

points_from – Source points (numpy array with shape [n, 2] where columns are [lon, lat])
points_to – Destination points (optional, defaults to points_from)

Returns:

Distance matrix as numpy array with shape [len(points_from), len(points_to)] Values are in meters.

External APIs¶

External API integrations for distance calculations (OSRM, Google Maps).

allocator.distances.external_apis.google_distance_matrix(X: ndarray, Y: ndarray | None = None, api_key: str | None = None, duration: bool = True) → ndarray[source]¶

Calculate distance matrix using Google Distance Matrix API.

Limitations:

Users of the standard API: * 2,500 free elements per day, calculated as the sum of client-side and

server-side queries.

Maximum of 25 origins or 25 destinations per request.
100 elements per request.
100 elements per second, calculated as the sum of client-side and server-side queries.

For more informaton:- https://developers.google.com/maps/documentation/distance-matrix/usage-limits

allocator.distances.external_apis.osrm_distance_matrix(X: ndarray, Y: ndarray | None = None, chunksize: int = 100, osrm_base_url: str | None = None) → ndarray[source]¶

Calculate distance matrix of arbitrary size using OSRM

Credits: https://github.com/stepankuzmin/distance-matrix

Please note that OSRM distance matrix is in duration in seconds.

Usage Examples¶

Basic Distance Matrix¶

import pandas as pd
import allocator

# Geographic locations
locations = pd.DataFrame({
    'longitude': [100.5, 100.6, 100.7],
    'latitude': [13.7, 13.8, 13.9],
})

# Calculate distance matrix
distances = allocator.get_distance_matrix(
    data=locations,
    method='haversine'
)

print(f"Distance matrix shape: {distances.shape}")
print(f"Distance from point 0 to point 1: {distances[0,1]:.2f}km")

Performance Comparison¶

import time

methods = ['euclidean', 'haversine']
for method in methods:
    start = time.time()
    distances = allocator.get_distance_matrix(locations, method=method)
    elapsed = time.time() - start
    print(f"{method}: {elapsed:.3f}s")

Distance Methods¶

Euclidean Distance¶

Method: euclidean
Formula: Straight-line distance in Cartesian coordinates
Speed: Fastest
Accuracy: Approximate for geographic coordinates
Best for: Quick estimates, large datasets, non-geographic data

# Fast approximate distances
distances = allocator.get_distance_matrix(data, method='euclidean')

Haversine Distance¶

Method: haversine
Formula: Great-circle distance on Earth’s surface
Speed: Fast
Accuracy: High for geographic coordinates
Best for: Most geographic applications (recommended)

# Accurate geographic distances
distances = allocator.get_distance_matrix(data, method='haversine')

OSRM Driving Distance¶

Method: osrm
Data source: OpenStreetMap road networks
Speed: Moderate (API calls)
Accuracy: Realistic driving distances
Requirements: Internet connection, OSRM server access

# Realistic driving distances
distances = allocator.get_distance_matrix(data, method='osrm')

Google Maps Distance¶

Method: googlemaps
Data source: Google Maps road data
Speed: Slow (API calls, rate limits)
Accuracy: High-quality driving distances
Requirements: Google Maps API key, billing account

# High-quality driving distances (requires API key)
distances = allocator.get_distance_matrix(
    data, 
    method='googlemaps',
    api_key='your_api_key'
)

API Configuration¶

OSRM Configuration¶

# Custom OSRM server
distances = allocator.get_distance_matrix(
    data,
    method='osrm',
    osrm_url='http://your-osrm-server:5000'
)

Google Maps Configuration¶

# Google Maps with API key
import os
os.environ['GOOGLEMAPS_API_KEY'] = 'your_api_key'

distances = allocator.get_distance_matrix(data, method='googlemaps')

Return Format¶

All distance methods return a NumPy array where:

distances[i,j] = distance from point i to point j
Diagonal elements are 0 (distance from point to itself)
Matrix is symmetric for undirected distances
Units are always kilometers

Performance Characteristics¶

Method	Speed	Accuracy	Use Case
euclidean	★★★★★	★★☆☆☆	Quick estimates, large datasets
haversine	★★★★☆	★★★★★	Geographic analysis (recommended)
osrm	★★★☆☆	★★★★☆	Realistic routing
googlemaps	★☆☆☆☆	★★★★★	Production routing applications

Integration with Other Functions¶

Distance methods are used automatically by other allocator functions:

# Clustering with specific distance method
clusters = allocator.cluster(data, distance='haversine')

# Routing with specific distance method  
route = allocator.shortest_path(data, distance='euclidean')

# Assignment with specific distance method
assignments = allocator.assign(points, workers, distance='osrm')

Error Handling¶

try:
    distances = allocator.get_distance_matrix(data, method='googlemaps')
except Exception as e:
    print(f"API error: {e}")
    # Fallback to local method
    distances = allocator.get_distance_matrix(data, method='haversine')

Advanced Usage¶

Custom Distance Functions¶

def custom_manhattan_distance(data):
    """Manhattan distance in kilometers."""
    # Convert to UTM coordinates for accurate metric calculation
    # ... custom implementation ...
    pass

# Use custom function with clustering
clusters = allocator.cluster(data, distance=custom_manhattan_distance)

Batch Processing¶

# Process large datasets in chunks
def batch_distance_matrix(data, chunk_size=100):
    """Calculate distance matrix in batches."""
    n = len(data)
    distances = np.zeros((n, n))
    
    for i in range(0, n, chunk_size):
        chunk = data.iloc[i:i+chunk_size]
        chunk_distances = allocator.get_distance_matrix(chunk)
        distances[i:i+len(chunk), i:i+len(chunk)] = chunk_distances
    
    return distances