Research on the Application of Machine Learning in Urban Public Transport Scheduling

Abstract

As a fundamental component of the urban transportation system, the frequency of bus departures significantly influences the commuting experience of residents along transit routes and the overall efficiency of urban ground transportation. However, in certain cities, challenges such as morning and evening rush-hour traffic, urban spatial planning around transit corridors, and continued reliance on manual dispatching systems hinder the timely and efficient scheduling of buses in accordance with real-time road conditions and passenger demand. On the basis of field observations of bus operations in City A, the author reported that during peak hours, long intervals between buses result in excessive passenger loads per vehicle, whereas during off-peak hours, low passenger demand leads to a high rate of underutilized seating capacity. This paper presents a comparative analysis between the traditional manual static dispatching method and an intelligent dispatching approach based on machine learning. Through a demonstration of the capabilities of machine learning in urban bus dispatching, the study highlights that a machine learning-integrated dispatching system offers superior performance in terms of route efficiency and dynamic adaptability to changing traffic conditions compared with conventional manual dispatching methods.