Estimating bus dwell time: A review of the literature

An accurate estimation of bus dwell time (BDT) is important as it directly influences the prediction of vehicles' arrival times at bus stops and hence, the reliability of their services. Developing an efficient method to estimate BDT has received significant attention in the literature as numerous studies have been conducted on this topic. This paper provides an overview of research works conducted in the past on bus dwell time. It covers three aspects: first methods used in the literature to estimate BDT, second factors influencing BDT and third methods used in the literature to collect data. The methods used in the literature to estimate BDT are broadly classified and discussed under three categories namely, regression-based models, probabilistic and time series models and artificial intelligence-based models along with their strengths and weaknesses. The influence of different policy measures related to characteristics of bus, passengers, time-of-day, transit, and route are discussed. The research trends reveal an increasing tendency to employ more advanced methods for data collection (using GPS, APC and AVL) and modelling (e.g. using machine learning-based methods).