Bank Customer Churn Prediction Using Machine Learning

Abstract

The banking sector is fiercely competitive in the present difficult time. Banks concentrate on both customer retention and customer turnover to raise the caliber and degree of service. The classification issue in the banking sector is examined in this essay. It detects possible churners from among potential customers and primarily focuses on bank customers' worries around churn. The bank uses supervised machine learning to identify and forecast which of its clients are most likely to leave. Since it is necessary to define churn and non-churn clients, customer churn prediction can be used in this situation. To address the distinctions between churn and non-churn clients, this study uses logistic regression, decision trees, and random forest classifiers. Accuracy levels can be attained via several classifiers. The Kaggle dataset for bank customer churn modeling is used for the experiment. To identify an appropriate model with more accuracy and predictability, the outcomes are compared. The findings demonstrate that, upon oversampling, in terms of accuracy, the decision tree model outperforms other models.