Machine Learning (ML) serves a crucial role in e-commerce datasets by providing valuable insights and predictive capabilities. This paper explores the realms of customer behavior prediction by employing machine learning techniques on e-commerce datasets focusing on Bank customer churn. The objective is to understand and predict why customers might leave a bank. In this paper, three Filter Feature selection algorithms namely Chi-Square, Information Gain, and Univariate Feature Selection in conjunction with classification algorithms such as Decision Tree (DT), Support Vector Machines (SVM), Random Forest, K-Nearest Neighbors (k-NN), Multilayer Perceptron (MLP), Logistic Regression, and Gaussian Naïve Bayes have been used. To determine the most effective and precise method to predict customer churn, the performance of these algorithms is also being assessed against a variety of performance metrics both with and without the feature selection methods. According to the research, Random Forest exhibits the highest accuracy of 99.91% for predicting customer attrition, and when the Chi-square feature selection method is used, it performs somewhat better with an accuracy of 99.95%.