Risk assessment method and device based on tree model feature derivation and electronic equipment

Abstract

The invention discloses a risk assessment method based on tree model feature derivation, which is used for assessing financial risks of a user and is characterized by comprising the following steps: acquiring application list information of a mobile terminal associated with the user and performing one-hot coding on the application list information to form one-hot coding features associated with the user; substituting the one-hot coding features into a tree feature derivation model to obtain one-hot coding derivation features associated with the user; constructing a machine learning model, andtraining the machine learning model by using the one-hot coding derivative features and financial performance data of the historical users; and inputting the one-hot coding derivative features of thenew user into the trained machine learning model, and calculating the risk score of the new user. The credit risk result evaluated by the method is high in accuracy and good in stability, and in the aspects of modeling and feature maintenance, less manpower is invested, resources are saved, and the effect is remarkable.

Description

technical field [0001] The present invention relates to the field of computer information processing, in particular to a risk assessment method, device, electronic equipment and computer-readable medium based on tree model feature derivation. Background technique [0002] In the field of consumer finance, under the premise that users are fully authorized, fully mining the user's application program (APP) data and using it for risk control can distinguish a certain proportion of high-risk users. At present, traditional application variable mining is mainly to classify and count applications, and then describe user portraits; or describe user risk changes according to changes in application installation and uninstallation in different periods; or derive relevant information based on the risk level of installed applications. into the model variable. The above methods can derive variables with a certain degree of risk differentiation, but these methods are generally difficult t...

AI Technical Summary

Problems solved by technology

The above methods can derive variables with a certain degree of risk differentiation, but these methods are generally difficult to maintain and require more manpower. The quality of derived variables depends to a large extent on the digger's understanding of the business, high maintenance costs, Unsatisfactory stability, etc.

Images