基于xgboost模型的消费者信用评级系统

Consumer Credit Rating System Based on the Xgboost Model

设计了一种基于xgboost模型的消费者信用评级系统,通过人脸识别方法选出一部分特征作为消费者标签,量化消费者信息,以此叙述消费者形象;对k-means聚类进行改进,提出了基于核密度的人脸识别聚类算法,将消费者分成不同的类别,据此完成信用评级。系统能够缓解噪声点敏感,使原始中心点选择更加简单,并且较少使用银行交易记录,具有较高的可用性。

A consumer credit rating system based on xgboost model is designed,which uses face recognition method to select some features as consumer labels,quantifies consumer information,and narrates consumer image.The k-means clustering is improved,and a face recognition clustering algorithm based on kernel density is proposed,which divides consumers into different categories for credit rating.The system can alleviate noise point sensitivity,make the selection of original center point more simple,and use less bank transaction records with high availability.