Predictive Analytics of Bus and Streetcar Bunching Occurrence Time for Real-Time Applications

过境车辆拥挤是过境机构常见和众所周知的运营问题。它对服务质量和效率以及用户感知都有负面影响。在研究和实践中，调查工作主要集中在公共汽车群聚，通过检查一系列因素的影响或制定纠正措施，以减轻群聚一旦发生的影响。这些研究中的大多数都是出于探索性目的而开发的。很少有研究关注有轨电车的聚束或开发实时应用的聚束预测分析。针对这一空白，本研究旨在开发公交和有轨电车线路沿着群聚发生时间的预测模型，丰富公交管理这一关键领域的知识和实践。正确预测实时聚束时间的能力将使运输机构能够采取先发制人的行动，防止聚束发生或至少减轻其影响。这只会提高服务效率和乘客满意度。本研究将使用多伦多的有轨电车和公共汽车系统作为案例研究，提供对多伦多地面网络的可靠性和效率有直接影响的现成输出。然而，研究输出预计将是通用的，足以转移到加拿大其他地方和地球仪的其他机构。

Transit vehicle bunching is a common and well-known operational problem for transit agencies. It has negativeimpacts on service quality and efficiency as well as on users' perception. In both research and practice,investigation efforts have mainly focused on bus bunching by examining the impacts of a range of factors or bydeveloping corrective actions to mitigate the effects of bunching once it occurs. Most of these studies weredeveloped for exploratory purposes. It is rare to find studies that focused on streetcar bunching or developedpredictive analytics of bunching for real-time applications. In response to this gap, this research aims atdeveloping predictive models of the time to bunching occurrence along bus and streetcar lines, enriching theknowledge and practice in this critical area of transit management. The ability to correctly predict the time tobunching in real-time would allow transit agencies to take pre-emptive actions towards preventing bunchingoccurrence or at least mitigating its effects. This can only result in improving the service efficiency and ridersatisfaction. This research will use Toronto's streetcar and bus systems as a case study, providing ready-to-useoutput that will have a direct impact on the reliability and efficiency of the surface network in Toronto.However, the research output is expected to be generic enough for transfer to other agencies elsewhere inCanada and across the globe.