| description abstract | At present, the green channel policy is fully implemented. However, the detection is relatively lagging, and truck drivers are susceptible to fake the toll-free logistics vehicles (TFLVs), causing huge losses to the operation. In order to improve the accuracy and efficiency of TFLVs inspection, this paper establishes a toll evasion prediction model for fake TFLVs based on the historical TFLVs traffic dataset derived from the highway network toll collection system. First, according to the importance and reliability, we used the data mining technology to differentiate and extract the data attributes. And the spatiotemporal characteristics and other characteristics of fake TFLVs through-traffic entering and exiting toll booths are studied and analyzed based on the preprocessed data. This study then used the Borderline-SMOTE oversampling method to balance the pass data set and the ChiMerge algorithm to discretize continuous attributes. In order to ensure the effective matching and the correlation between the large contribution attributes and the results, correlation item test and collinearity test are used for discrete related attributes. Finally, adopting the processed discrete TFLVs data which passed correlation item test and collinearity test, and using the decision tree to establish the prediction model, the classification results of the escape behavior prediction model and other models are compared by using the historical TFLVs data set. The results show that the accuracy of the escape behavior prediction model proposed in this paper is 83.4%, which is higher than that of the Logistic regression model (61.8%) and the Random forest model (81%). This research model can effectively warn of fake TFLVs, and on the basis of simplifying the inspection process of TFLVs, can reduce the probability of the occurrence of fake TFLVs evasion, which has practical application significance. | |
