ProjetAI — Algiers Multi-Transport Transit Network

Multi-modal transit dataset + UCS algorithm for route planning across bus, metro, tram, train, and télépherique in Algiers.

Quick Start

from ucs import TransitRouter

router = TransitRouter('data/')
result = router.find_route('M1_MARTYRS', 'TR01', metric='time')

print(f"Time: {result.total_time} min")
print(f"Distance: {result.total_dist} km")
print(f"Fare: {result.total_fare} DA")

for seg in result.segments:
    print(f"  {seg.transport_type.upper()}: {seg.from_stop} → {seg.to_stop} ({seg.time_min} min)")

Dataset

All files in data/:

File	Description
stops.csv	1,314 stops (nodes) with coordinates
edges.csv	19,139 directed edges (the graph)
routes.csv	228 routes/lines
transfers.csv	11,208 walking/transfer edges
graph_adjacency.json	Adjacency list — ready for any algorithm
graph_edge_list.csv	Simple edge list format
transit_graph.graphml	GraphML — for NetworkX
bus_geometries.json	GPS road geometry for bus paths (map rendering)
schema.sql	SQL CREATE TABLE statements
inserts.sql	SQL INSERT statements

Graph Structure

• Nodes: 1,314 stops across 5 transport types
• Edges: 19,139 directed weighted connections
• Fully connected: 100% of nodes reachable
• Edge weights: distance_km and time_min

Loading the graph

import json

with open('data/graph_adjacency.json') as f:
    graph = json.load(f)

# graph["M1_MARTYRS"] → [{"to": "M1_BOUMEN", "dist": 2.317, "time": 3.0, "type": "metro", "route": "METRO_L1"}, ...]

For NetworkX

import networkx as nx
G = nx.read_graphml('data/transit_graph.graphml')
path = nx.shortest_path(G, 'M1_MARTYRS', 'TR35', weight='time_min')

UCS Algorithm (ucs.py)

• Binary heap priority queue
• Predecessor-based path reconstruction
• Fare calculation: bus charges only on line change
• Nearest-stop finder by coordinates
• Supports time and distance optimization

CLI Mode

python ucs.py data/

Fare Table (DA)

Mode	Fare
Bus	30 per line (only on line change)
Metro	50
Tram	40
Train	50
Télépherique	30
Walk	0

Bus Road Geometry

bus_geometries.json contains real GPS waypoints for bus edges (from 61K trace points). Use for map rendering:

import json
geom = json.load(open('data/bus_geometries.json'))
# Key format: "from_stop_id|to_stop_id|route_id"
# Value: [[lat, lon], [lat, lon], ...] — road waypoints