Supervised learning-based high-quality customer loss prediction method and device, and storage medium

Abstract

The invention discloses a supervised learning-based high-quality customer loss prediction method and device, and a storage medium. The method comprises the steps of: acquiring a to-be-lost predicted high-quality customer, and extracting loss feature data of the to-be-lost predicted high-quality customer in a preset time period; inputting the loss feature data of the to-be-lost predicted high-quality customer into a pre-trained high-quality customer loss prediction model, and outputting a high-quality customer loss prediction result; wherein the high-quality customer loss prediction model is obtained by training a prediction model constructed based on an XGBoost algorithm by using a constructed high-quality customer loss prediction sample data set. A high-quality customer group is selectedas the research object, so that the category imbalance condition of training data is reduced, the training speed and accuracy of the model are improved, and core customers of a bank can be mastered; an XGBoost algorithm is used, the algorithm is not sensitive to missing values, and interpolation processing does not need to be carried out on the missing values; the learning ability is high, the training speed is high, and the convergence speed is high.

Description

technical field [0001] The invention relates to the technical field of customer loss prediction of commercial banks, in particular to a method, device and storage medium for high-quality customer loss prediction based on supervised learning. Background technique [0002] The current market competition is becoming increasingly fierce, and the product or service differences between banks are getting smaller and smaller. More and more banks are shifting from "product-centric" to "customer-centric", and adopt information technology such as customer relationship management (CRM) one after another. System to improve system service level, customer churn is the focus of CRM and one of the core issues of the banking industry. [0003] Relevant studies have found that the cost of winning a new customer is 5 to 6 times that of retaining an old customer. Therefore, it is of strategic significance to improve the competitiveness of enterprises to retain old customers, predict potential l...

AI Technical Summary

Problems solved by technology

[0005] Each of the above steps determines the effect of the churn prediction model in practical applications: it is often not convincing to define churn by using the action of canceling cards and accounts. For example, some customers have lost contact with the bank, but there may not be sales. Therefore, under this definition, the scope of the target group is often reduced; it is often inaccurate to select a single product for churn judgment. For example, although a customer reduces or cancels a fixed deposit, but purchases a corresponding or more amount At this time, using the single-product cancellation to judge the loss of investment products will cause the definition of loss to be inconsistent with the reality; at the same time, using the characteristics that do not meet the basic assumptions of the model as input will lead to wrong conclusions, such as the Cox proportional hazard model The variables required to be used are not related to time, etc.; finally, the learning ability of the model also determines the final prediction effect

Images