Proceedings of the 4th Management Science Informatization and Economic Innovation Development Conference, MSIEID 2022, December 9-11, 2022, Chongqing, China

Research Article

Bank Customer Churn Analysis and Prediction

Download800 downloads
  • @INPROCEEDINGS{10.4108/eai.9-12-2022.2327608,
        author={Wenhui  Zhang},
        title={Bank Customer Churn Analysis and Prediction},
        proceedings={Proceedings of the 4th Management Science Informatization and Economic Innovation Development Conference, MSIEID 2022, December 9-11, 2022, Chongqing, China},
        publisher={EAI},
        proceedings_a={MSIEID},
        year={2023},
        month={3},
        keywords={banking customer churn crosstabs analysis independent-samples t test factor analysis one-way anova logistic regression},
        doi={10.4108/eai.9-12-2022.2327608}
    }
    
  • Wenhui Zhang
    Year: 2023
    Bank Customer Churn Analysis and Prediction
    MSIEID
    EAI
    DOI: 10.4108/eai.9-12-2022.2327608
Wenhui Zhang1,*
  • 1: Beijing Normal Unversity
*Contact email: 201911030129@mail.bnu.edu.cn

Abstract

The term "Customer Churn" refers to the state in which the customer stops using products or services of a company. Although the bank will inevitably lose users, which is inevitable in the process of replacing the old and new banking users, the proportion and changing trend of lost users can indicate the bank's ability to retain users and the development trend of the bank. Therefore, it is necessary for banks to know the reasons leading a client to leave the company. In order to explore the factors affecting the loss of bank users, this paper selects a dataset obtained from Kaggle, using the methods of crosstabs analysis, independent samples T test, factor analysis and one-way ANOVA respectively. In addition, logistic regression is also used to predict customer churn. From the t-test, this paper finds that those who quit have lower credit scores, are older and have larger balances than those who don't. From the factor analysis, this paper finds that the feature of country and balance are the most explanatory factors. From the logistic regression, this paper finds that the percentages of correct predictions are 69.5% and 65.3% respectively before and after selecting main components.