Details
Abstract
Overdue risk detection is a critical issue faced by consumer finance companies. As a high overdue rate will result in economic losses, it is of significance to detect risky customers. However, the large volume of credit data makes manual expert analysis rather challenging. Additionally, previous machine learning approaches neglect the relations between different customers, and traditional graph neural networks lack the exploration of overdue patterns. In this paper, we construct a heterogeneous graph based on real credit data. We propose a meta-path-based graph neural network and the experimental results show that our model achieves the best effect in our task.