Loan risk timeliness prediction system and method based on LSTM

Abstract

The invention provides a loan risk timeliness prediction system based on LSTM. The loan risk timeliness prediction system comprises a server storage module, a long and short memory LSTM module, a strong learner GBDT training module, a parameter storage module and a server selection module which are connected in sequence. Based on the above system, the invention also discloses a loan risk timeliness prediction method based on LSTM. According to the invention, user timeliness data is combined; statistical analysis is performed on personal fixed assets, identity information and behavior characteristics of the borrowing and lending users; LSTM and GBDT are used to carry out regression prediction; therefore, on the basis of conforming to the personal attributes and the behavior characteristicsof the user, analysis can be performed according to the personal information and the behavior history of the user, and the probability of possible risks during re-borrowing is predicted, so that a result of recommending the lending transaction to a financial institution is provided, and the lending risk is reduced.

Description

technical field [0001] The invention relates to the technical field of computer data processing, in particular to an LSTM-based system and method for predicting the timeliness of lending risk. Background technique [0002] As e-commerce brings convenience to everyone without leaving home, mobile payment has also become a mainstream payment platform. At the same time, major lending platforms and mobile payment have become common payment methods for people, such as borrowing components bundled with payment platforms, specialized apps, and credit cards. However, for financial institutions, while lending money to earn interest, it is inevitable that users cannot pay off the arrears within the specified time, resulting in high recurring interest and causing various financial disputes. Therefore, borrowing users Whether the loan can be paid off within the specified time, and how much loan amount should be set for the user has become a crucial issue. [0003] During the operation...

AI Technical Summary

Problems solved by technology

[0004] There has been a lot of related work on loan risk prediction methods, but the existing methods are all based on constructing a static user portrait model, which needs to be constructed based on static user portrait feature engineering, namely: (1) The user attributes corresponding to the portrait method are fixed; (2) the total amount of user loans corresponding to the platform is fixed; in reality, it is difficult to ensure that the user's attributes remain unchanged, or the user's social information does not change, so It greatly reduces the control of lending risk

The problems caused by this are: (1) Changes in the user's occupation or fixed assets or changes in the social circle will have a greater impact on personal finances, and it is impossible to reflect the changes in repayment behavior caused by such impacts in a timely manner

(2) The user repays the loan in advance due to uncertain factors, fails to repay the loan in time, or cannot repay the full amount. Although the existing methods take this factor into account, they only use feature engineering for fixed analysis, and do not consider Time factor, the accuracy of loan risk prediction is low

Images