Short-term Traffic Flow Prediction based on Data-Driven Knearest neighbour Nonparametric Regression

• Published in: Journal of physics. Conference series Vol. 1213; no. 5; pp. 52070 - 52075
• Main Authors: Chen, Ken, Zhao, Shasha, Zhang, Dengyin
• Format: Journal Article
• Language: English
• Published: Bristol IOP Publishing 01.06.2019
• Subjects: Algorithms; Data search; K-nearest neighbors algorithm; Nonparametric statistics; Prediction models; Predictions; Regression models; Search algorithms; Traffic flow; Traffic models
• ISSN: 1742-6588; 1742-6596
• DOI: 10.1088/1742-6596/1213/5/052070

In order to solve the problem that the KNN algorithm is not accurate enough in the short-term traffic forecasting process and the prediction efficiency is not high caused by the search for past observations, a short-term traffic flow prediction model based on data-driven K-nearest neighbor non-parametric regression is proposed. The model is developed based on two-step search algorithm. Firstly, in the non-predictive period of time, the candidate input data similar to the current state is searched from the historical data for the system. Then, the optimal decision input data for prediction is identified from the candidate input data at the predictive point. Finally, the best decision input data is used to generate the prediction through the prediction algorithm. The simulation results show that the algorithm can effectively reduce the time for searching historical data on the premise of ensuring the accuracy of system prediction, thus reducing the execution time in the process of system prediction and improving the prediction efficiency of the system.