Prediction of bus arrival time to stations by using AVL data: Case study of Qazvin public bus system

City bus transportation system is considered as one of the most important transportation infrastructure system for transporting passengers in world. One solution to increase the quality of bus services and to meet passengers’ needs is to change their attitude about bus service and make a shift from private transportation to pubic mobility. In recent years, Automatic vehicle location (AVL) systems have been used in order to improve transportation services quality in Iran, but mostly the gathered data are not analyzed for gaining customers satisfaction. Given the fact that knowing exact bus arrival time to stations is one of the most important needs of passengers, in this survey by using Qazvin city bus system datasets, which is mainly based on travel time and headways, a prediction model is presented. The model uses AVL data and some basic well-known statistical and machine learning models such as Neural Networks, ARIMA time series and Support Vector Machines. According to our results, ANN model performs better than other models in predicting bus arrival time to stations. Given that similar research has not been done to predict the arrival time of buses in Iran, This article has attempted to use the well-known techniques and real-world data, provide empirical view in this scope.