Research On Default Of Carrier Of NTOCC Based On Data Mining

As a new format in the transportation field,the network freight platform has solved the problem of high logistics costs in my country's road freight market to a certain extent.The online freight platform adheres to the "exchange transaction" model.For each order,the platform will directly sign an electronic contract with the carrier and the shipper,and then entrust the carrier to transport the goods.However,in the process from the carrier's pick-up and transportation to the settlement of the receipt,there are situations in which the order cannot be completed due to the carrier's default,which damages the shipper's rights and interests and affects the shipper's satisfaction and stickiness with the platform.Even some carriers breached the contract after receiving the prepaid freight,but did not refund the freight,which caused direct economic losses to the platform.At present,the development of my country's online freight platform is not yet mature.The above-mentioned risk management and control issues are issues that must be overcome on the road to sustainable development of the online freight platform.This article takes the network freight platform carrier's breach of contract as the research object,and has done the following work: First,by drawing on the research results of the user's breach of contract in the fields of finance and power,and analyzing the characteristics of the network freight platform and the carrier,the network The connotation and manifestation of the freight platform carrier's breach of contract are defined;second,considering that the platform should take different measures against the breach of contract by carriers of different values,this article derives the attributes of the traditional user value model and constructs the integration of user risk Based on the carrier value segmentation model of the network freight platform,and based on this model,the K-means algorithm is used to achieve carrier category segmentation,and obtain high-risk high-value carriers,medium-risk low-value carriers,and low-risk medium-value carriers,and then According to the characteristics of sub-user groups,provide corresponding operation strategies and risk control suggestions for the scientific management of the carrier on the network freight platform;finally,in order to realize the early warning of the carrier's breach of contract,this paper uses the carrier segmentation result obtained in the previous step as the characteristic factor,And combined the characteristics of the carrier's historical behavior,order information,vehicle information and other dimensions to construct a carrier default prediction index system,and then train a better performance of random forest,XGBoost,Light GBM classification models.Through the analysis of the actual data values of the M platform,the result labels of the carrier value segmentation results have a positive effect on the prediction model.Furthermore,in order to obtain a better predictive model,this paper uses the Stacking model fusion method to fuse the above models,and obtains a predictive model with a recall rate of 98.45%,and explains the application scenarios of the model for different types of carriers.The research in this article has important significance and reference value for the network freight platform to timely detect the carrier's breach of contract,take correct customer relationship management measures,and further reduce platform risks,thereby improving user satisfaction and stickiness,and improving corporate profits.