Train Delay Predicti On Using Machine Learning and Deep Learning Techniques

Abstract

Transport systems are critical pieces of infrastructure and they have substantially increased in size in many countries worldwide. This includes rail transport systems that have evolved significantly, including to provide long- distance travel services. Passenger train delay significantly influences riders’ decision to choose rail transport as their mode choice. Poor on-time performance can impact passenger trust and their satisfaction, and it may result in a shift to other modes of transport, especially private vehicles and air transport. Service disruption is a root cause of lower rail punctuality and customer satisfaction. Major service disruptions result from various conditions or factors such as accidents, problems in train operation, malfunctioning or damaged equipment, routine maintenance, construction, passenger boarding or alighting, and even extreme weather conditions. Train delay can also negatively affect connecting trains and passengers’ journeys or activities. Thus, delay estimates or predictions can help train operators develop better plans to manage, reschedule, or adjust the timetable of the current and consecutive trains more effectively, as well as to inform passengers in advance so they themselves can adjust their travel plans in time. In light of these problems, the main objective is to model passenger train delay prediction based on three Machine Learning and Deep Learning techniques.