Congestion Prediction on Rapid Transit System Based on Weighted Resample Deep Neural Network
Investigating congestion in train rapid transit system (RTS) in today’s urban is demanded by both the operators and the public. Increase traffic data availability can be obtained from travel smart card and allowed to investigate the congestion of RTS. Artificial neural network are employed to do prediction on traffic. However the imbalance of data is a challenge to make an efficient prediction on congestion of RTS. This work proposes a Weighted Resample Deep Neural Network (WRDNN) model to predict the congestion level of RTS. The case study of RTS of one city of US indicate that the model introduced in this work can effectively predicting the congestion level of RTS with the 90% accuracy..
KeywordsCongestion prediction Deep neural networks Data imbalance Rapid transit system
This research is funded by Fujian Provincial Department of Science and Technology (Granted No. 2017J01729) and the China Scholarship Council.
- 3.Sun, L., Lee, D.-H., Erath, A., Huang, X.: Using smart card data to extract passenger’s spatio-temporal density and train’s trajectory of MRT system. In: Proceedings of the ACM SIGKDD International Workshop on Urban Computing, pp. 142–148. ACM (2012)Google Scholar
- 13.Powers, D.M.: Evaluation: from precision, recall and F-measure to ROC, informedness, markedness and correlation (2011)Google Scholar
- 14.Hu, R., Xia, Y.: Traffic condition recognition based on vehicle trajectory big data. J. Internet Technol. 18(7), 1587–1596 (2017)Google Scholar