Dynamic spatial-temporal feature optimization with ERI big data for Short-term traffic flow prediction

Abstract

Accurate short-term traffic flow prediction is an important basis of intelligent transportation systems (ITS) such as transportation operations and urban planning applications. However, due to the lack of complete directly measured data on urban traffic flow, existing studies cannot adequately mine the dynamic spatial-temporal correlations characterizing traffic flows in urban road networks. Electronic registration identification (ERI), which is an emerging technology for uniquely identifying a vehicle, can help collect the travel records of all vehicles. This inspires us to employ ERI big data for traffic flow prediction. In this paper, we propose a dynamic spatial-temporal feature optimization method with ERI big data for short-term traffic flow prediction based on a gradient–boosted regression tree, called DSTO-GBRT. Firstly, the framework of DSTO-GBRT is built. Secondly, we analyze the dynamic spatial-temporal correlations among the current prediction point and upstream correlative points using the Pearson correlation coefficient (PCC). Thirdly, to eliminate the linear correlations among features, we exploit principal component analysis (PCA) to optimize the original training data and obtain optimized training data. In the experiment, real-world ERI big data from Chongqing are employed for the proposed DSTO-GBRT method. Compared with ST-GBRT, ARIMA, DSTO-BPNN and DSTO-SVM, the results demonstrate that DSTO-GBRT can provide timely and adaptive prediction even in rush hour, when traffic conditions change rapidly. Furthermore, compared with DSO-GBRT and DTO-GBRT, the results show that the proposed DSTO-GBRT method is more accurate.