Research Article
Traffic Flow Estimation for Urban Roads Based on Crowdsourced Data and Machine Learning Principles
@INPROCEEDINGS{10.1007/978-3-319-93710-6_27, author={Sakitha Kumarage and R. Rajapaksha and Dimantha Silva and J. Bandara}, title={Traffic Flow Estimation for Urban Roads Based on Crowdsourced Data and Machine Learning Principles}, proceedings={Intelligent Transport Systems -- From Research and Development to the Market Uptake. First International Conference, INTSYS 2017, Hyvink\aa{}\aa{}, Finland, November 29-30, 2017, Proceedings}, proceedings_a={INTSYS}, year={2018}, month={7}, keywords={Traffic flow estimation Machine learning KNN regression Google APIs Crowdsourced data}, doi={10.1007/978-3-319-93710-6_27} }- Sakitha Kumarage
R. Rajapaksha
Dimantha Silva
J. Bandara
Year: 2018
Traffic Flow Estimation for Urban Roads Based on Crowdsourced Data and Machine Learning Principles
INTSYS
Springer
DOI: 10.1007/978-3-319-93710-6_27
Abstract
The congestion in urban road networks are common problem across all urban centers. Understanding the traffic flow across the road segments are necessary to provide viable solutions, but a very expensive task specially for developing countries. This paper proposes an economical approach for a directional flow prediction model for urban road based on Google Distance Matrix API data, archived traffic flow data, and geometric data. Data gathered was aggregated in space and time as attributes to the model estimation. Deviating from traditional probability estimation, a K- Nearest Neighbour regression method was used in the analysis. The model is validated using a test dataset which showed a root mean square error and a mean absolute error of prediction as 9.479 and 2.318, which suggest that with the use of travel time and speed data gathered from Google Distance matrix API is possible to estimate lane flow when road geometry is defined.