A study on Ant Colony-XGBoost based business reliability portrait method for network freight transportation enterprises

In recent years, the network freight industry in China has been developing rapidly, and has become an important force in increasing efficiency in logistics, but there is still a lack of effective portrait methods for network freight enterprises in the society. This paper combines the database of network freight transport regulatory system of the Chinese Ministry of Transport, proposes the XGBoost model portrait label classification algorithm based on ant colony optimisation, and researches the research on the portrait of network freight enterprises by taking a certain province in the system as an example. By comparing with the ranking of network freight transport enterprises released by China Federation of Logistics & Purchasing (CFLP), the reasonableness of network freight transport enterprise portrait is verified, and comparing with logistic regression, random forest, KNN and other methods, the label classification accuracy of enterprise portrait method based on ACO-XGBoost reaches 95.71%, and the overall performance of the model is optimal. And the sensitivity analysis of the influence of business reliability characteristics is conducted, and it is determined that the normal rate of overload regulation, the normal rate of vehicle track and the driver qualification compliance rate are important factors affecting the enterprise portrait of network freight transport. The result proves the new algorithm can quickly and accurately find the optimal solution for profiling network freight enterprises.