Bus Pass Time Estimation based on Efficient Data Gathering from a Slow Mobility Server

Statistical description and prediction of bus arrival times is relevant for public transport users since it allows more timewise efficient journeys. This work is focused on characterizing the real behavior of buses based on past arrival estimation data. The main goal is to estimate real bus pass times by optimally collecting data from an intercity bus arrival time estimation system which is limited in petition handling capacity. This requires to model the server behavior prior to the design of the data collection system. In addition, it also requires the design of an algorithm to estimate the bus real passing time considering that only the provided estimated time of arrival is available. This information can be useful for designing alternative online arrival time estimators based on supervised learning which could potentially improve the estimator efficiency.