IVY-RF-Based Logistics Claims Risk Classification and Prediction

Abstract

With the rapid development of e-commerce, the scale and complexity of logistics operations continue to increase, and the claims risks caused by abnormal events such as cargo damage, delays, and loss during transportation are becoming increasingly prominent. The traditional claims processing model, primarily based on manual review, is no longer adequate for the high-frequency, large-scale business demands in terms of processing efficiency, decision consistency, and cost control. There is an urgent need to introduce intelligent methods to achieve accurate identification and hierarchical management of claims risks. Addressing the challenges of diverse feature dimensions, highly imbalanced category distribution, and difficulty in distinguishing different risk types in logistics claims data, this paper proposes a Random Forest Logistics Claims Risk Classification Model (IVY-RF) based on the IVY growth optimization algorithm. This method uses a random forest as the basic classifier, fully leveraging its advantages in nonlinear relationship modeling and feature interaction capture. It also introduces the IVY metaheuristic optimization algorithm to adaptively optimize the model’s key hyperparameters globally. Experimental results based on real-world logistics claims datasets demonstrate that the IVY-RF model significantly outperforms comparable models such as IVY-LightGBM and IVY-XGBoost in core evaluation metrics, including macro-average F1 score, weighted precision, and weighted recall, achieving a better performance balance between the majority and minority high-risk categories. The findings indicate that the proposed IVY-RF model exhibits significant advantages in prediction accuracy, stability, and engineering feasibility, providing reliable technical support for logistics companies to conduct intelligent identification and refined management of claims risks.