Data processing method and data processing device

Abstract

The invention provides a data processing method and a data processing device, relates to the field of artificial intelligence, and improves the accuracy of default prediction. The method comprises the following steps: acquiring transaction data to be predicted, wherein the transaction data to be predicted comprises data of at least one transaction to be predicted; inputting the to-be-predicted transaction data into a default prediction model, and outputting the default probability of the at least one to-be-predicted transaction in a future time period, the default prediction model being obtained by training a time convolution network (TCN) model by using historical transaction data; and judging a default risk level of the at least one piece of transaction data to be predicted based on the default probability.

Description

technical field [0001] The present application relates to the field of artificial intelligence, and in particular to a data processing method and a data processing device. Background technique [0002] As the total amount of credit in the credit card business continues to rise, banks continue to increase capital investment, and investment channels continue to expand, the risk of credit card business is worthy of attention. Although the increase in the total amount of credit card credit can bring more income to banks, the increase in the total amount of overdue credit and the total amount of credit payable at the end of the period will undoubtedly increase the probability of credit card risks. It is very difficult for the bank to monitor the card usage of each cardholder (ie user) in real time. Once a problem occurs, the interests of the bank will be damaged, and even systemic risks will be caused. [0003] At present, the quantitative assessment of credit risk uses a non-de...

AI Technical Summary

Problems solved by technology

Although the increase in the total amount of credit card credit can bring more income to the bank, the increase in the total amount of overdue credit and the total amount of credit payable at the end of the period will undoubtedly increase the probability of credit card risks.

It is very difficult for the bank to monitor the card usage of each cardholder (ie user) in real time. Once a problem occurs, the interests of the bank will be damaged, and even systemic risks will be caused.

[0003] At present, the quantitative assessment of credit risk uses a non-deep learning logistic regression model, but since the logistic regression model is a linear regression model, it is difficult to fit the real distribution of data, resulting in the problem of low efficiency and accuracy of default prediction

Images