Enhanced Bank Marketing System Using Optimized GSA-BP Neural Network for Improved Customer Targeting

Abstract

To more accurately predict the needs of bank customers and improve the marketing success rate of bank wealth management products. Study designed a bank marketing system that can mine potential customers by analyzing customer data. The discovery of potential customers is mainly achieved by constructing the BA GSA-BP model, which optimizes the structure of the backpropagation neural network through gravity algorithm and integrates the optimized neural network using Bagging set learning algorithm. The study conducted simulation tests on the performance of the model. In the Bank Marketing dataset, after 225 iterations, the MES of GSA-BP remained stable at 0.098; In the Credit Card Default dataset, after 164 iterations, the MES of GSA-BP remained stable at 0.093, In terms of accuracy and recall, GSA-BP also performs well. For example, in the PR curve of the Bank Marketing dataset, the recall rate is about 0.896 when the accuracy is 0.8, and about 0.800 when the accuracy is 0.9. After using the Bagging ensemble algorithm, the performance of GSA-BP was further improved, with the largest area under the ROC curve, indicating strong generalization ability and low risk of overfitting. When the number of users is 200, the response time of the research system is 0.24 seconds, and the prediction success rate is about 60%. The above data illustrates that the method used in the study improves the over-fitting problem of traditional BP neural network and enhances the generalization ability of the algorithm. The constructed bank marketing system has the characteristics of short response time and high prediction success rate. The results of this study can accurately predict whether customers will order bank promotional products, conduct targeted marketing for customers, reduce marketing risks, and improve the success rate of product marketing and provide reliable technical support for predictions in other industries.