Heuristic Bus Route Optimization Incorporating Taxi-Derived Route Preferences under NYC Congestion Pricing

This study proposes a data-driven optimization framework for designing efficient bus routes in Manhattan under the new congestion toll policy. Using origin-destination data from the New York City Yellow Taxi Trip Records, urban mobility is modeled as a directed graph, and a HybridScore function is defined to integrate congestion surcharges, Central Business District (CBD) fees, and trip frequency. Three metaheuristic algorithms, Genetic Algorithm (GA), Simulated Annealing (SA), and Ant Colony Optimization (ACO), are applied to maximize the weighted sum over a single K-cycle, representing a feasible bus loop. Comparative experiments at K = 10, 25, and 50 show that ACO consistently achieves the best performance, balancing solution quality, convergence speed, and stability, while SA rapidly produces feasible initial solutions. Beyond algorithmic performance, the results demonstrate the framework’s potential to inform real-world transit design by identifying corridors where expanded bus services could reduce congestion within Manhattan’s toll zone. This approach links computational optimization to sustainable policy implementation, offering a scalable method for building adaptive public transportation networks in dense urban environments.