Using support vector machine models for real-time crash risk prediction on urban expressways

摘要

This paper adopted a novel methodology—support vector machine (SVM) with two penalty parameters for the evaluation of real-time crash risk on urban expressway segments using the dual-loop detector data. The purpose of this study is to develop a model that can identify traffic conditions prone to crashes effectively and support implementation of proactive traffic safety management. Based on the crash data and the corresponding detector data collected on expressways of Shanghai, different combinations of dual-loop detector data and time segments before crashes were used to develop the optimal crash risk estimation model by SVM. Then, the transferability of SVM model was assessed by examining whether the model developed on one expressway is applicable to other similar ones. In addition, the prediction results and transferability of SVM model were compared with those given by other frequently used classification algorithms, including Logistic Regression, Bayesian Networks, Native Bayes classifier, K-Nearest Neighbor, and Back Propagation Neural Network. The results showed that SVM model is more suitable to the prediction of real-time crash risk with small-scale data than other algorithms with the crash classification accuracy reaching 80% at best. A comparison to the similar studies by other researchers also implied that the proposed model achieved better predication accuracy.