Travel time prediction using historical big open data

Abstract

The need for travel time estimations and prediction for both transit companies and travelers are increasing. Intelligent transportation systems are often plagued by a shortage of data sources to properly assess the traffic situation. This thesis propose an approach to improve the reliability of travel time predictions through the creation of a combined model that relies on traffic estimations from both buses and cars. We found that the use of multiple sources of traffic data can improve the accuracy and reliability of travel time estimations and prediction where one of the initial datasets suffer from data sparcity.