Bank credit risk prediction method and device

Abstract

The invention discloses a bank credit risk prediction method and device, and the method comprises the steps: carrying out the word segmentation of transaction flow related data, and determining the transaction information of different customers, wherein the transaction information comprises a transaction merchant and a merchant category code; according to the transaction information of different clients, establishing a transaction heterogeneous graph by taking each client and merchant as nodes; and determining a credit risk prediction value of each client according to the heterogeneous graph of the transaction by adopting a heterogeneous graph neural network algorithm. The transaction amount, frequency and channel between each client and different merchants are displayed through the heterogeneous graph, compared with the prior art, the problem that all transaction information related to the client cannot be integrated in the prior art is solved, the purpose of comprehensively analyzingall transaction information related to the client is achieved, and the accuracy of carrying out credit risk prediction on the client is improved. Actual transaction merchants and merchant category codes in transaction flow related data can be determined, and the applicability of bank credit risk prediction is improved.

Description

technical field [0001] The invention relates to the technical field of neural networks, in particular to a bank credit risk prediction method and device. Background technique [0002] This section is intended to provide a background or context to embodiments of the invention that are recited in the claims. The descriptions herein are not admitted to be prior art by inclusion in this section. [0003] The current customer credit risk prediction method is mainly based on deep learning and machine learning, etc., and is obtained through data mining of customer information, financial attributes, credit information and consumer behavior data. The credit risk forecasting method under the existing technology can only perform simple statistics on transaction flow-related data, and cannot consider transaction merchants and merchant categories. For example, if two same customers spend N sums of M yuan, but customer A buys home appliances and customer B buys stocks, the credit risks ...

AI Technical Summary

Problems solved by technology

Due to the existence of third-party payment platforms and online e-commerce platforms, customers do not directly use bank cards to trade with merchants, which makes it impossible for banks to identify the actual transaction merchants and merchant categories.

For example, in many online payments and offline consumption, in the bank's transaction flow related data, the transaction merchants correspond to the e-commerce platforms, such as the transaction merchants of all transaction records of bank cards bound to WeChat payment are Tencent payment, not specific consumption. The actual trading merchants such as Wal-Mart, etc., traditional technology cannot distinguish this

Therefore, when the transaction is paid through online channels, the transaction merchants connected with the bank are third-party payment channels or e-commerce platforms. Traditional credit risk prediction methods cannot accurately analyze the transaction flow related data of customers' online transactions, and cannot be accurate. Therefore, the traditional credit risk prediction method cannot accurately analyze online transactions, which has certain limitations and reduces the applicability of bank credit risk prediction

Images