Bus arrival time prediction model based on APC data

Bus arrival times are influenced by stochastic variations in number of factors, (e.g., intersection delays, traffic congestion, and weather conditions) resulting in buses to deviate from the predetermined schedule and lengthening of passenger waiting times for buses. Providing passengers with accurate bus arrival times through advanced traveler information systems can help passengers' decision making and reduce average waiting time for buses. In this paper, a set of models, predicting bus arrival times at stops along a route, are developed through mining historical data collected by automated passenger counters (APC). The data applied for developing the models were collected by APC installed on buses operated by the public transit agency in Harbin, a city of northeast region of China. The results obtained are promising, and indicate that the developed models can be used to estimate bus arrival times.