T2NBS: Planning Night-Time Demand-Oriented Bus Systems with Urban Computing Approaches

This paper has established an urban computing framework T2NBS (Taxi to Night-time Bus Service) in order to realize the optimal NBS (Night-time Bus Service) system. The framework includes (1) a reasonable mathematical model MNL for travel demand estimation and a multi-variable decision target MINLP for the optimal system output, which simultaneously optimize the NBS system’s design quality (site, path, schedule, vehicle-to-passenger matching) and service quality (walking accessibility, demand-matching degree, travel time and travel cost, the profit of the whole system); and (2) a staged sequential heuristic algorithm for solving the programming model’s “super integrated NP-Hard” problem, including the sub-algorithm for mining the night-time collective travel demands from taxi trajectories at the first stage, the second-stage deployment sub-algorithm for extracting the fewest sites with best walking accessibility, and the third-stage sub-algorithm for searching the best operation timetable with maximized revenues.