A Gaussian process regression method for urban road travel time prediction
Authors:
- 1 July 2017
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 890-894
Abstract
Urban road travel time prediction has become a crucial component in Intelligent Transportation Systems (ITS). In order to improve the overall operational efficiency of the urban expressway network, this paper proposes a Gaussian Process Regression (GPR) based prediction method to estimate the short period travel time on urban road. Compared with the approaches based on neural network and support vector machine regression model, GPR is easier to be implemented. Moreover, the proposed method is able to adapt to the urban road environment and has the advantages of hyper-parameter self-adaptive acquisition. Based on people's regular travel habits, this method considers the number of vehicles passing through the road section, the average vehicle speed, and the travel time jointly to predict the travel time of the next time period. We then predict the traveling time on the workdays and the weekends with the proposed method respectively. Results show that the prediction made by the proposed method is consistent with the actual result with 2.8% mean absolute percentage error and 97.2% prediction accuracy.Keywords
This publication has 3 references indexed in Scilit:
- A gradient boosting method to improve travel time predictionTransportation Research Part C: Emerging Technologies, 2015
- Improving the accuracy of long-term travel time prediction using heterogeneous ensemblesNeurocomputing, 2015
- Spectral Basis Neural Networks for Real-Time Travel Time ForecastingJournal of Transportation Engineering, 1999