Businesses depend on banks for financing and other services. The success or failure of a company depends in large part on the ability of the industry to identify credit risk. As a result, banks must analyze whether or not a loan application will default in the future. To evaluate if a loan application was eligible for one, financial firms used highly competent personnel in the past. Machine learning algorithms and neural networks have been used to train class-sifters to forecast an individual's credit score based on their prior credit history, preventing loans from being provided to individuals who have failed on their obligations but these machine learning approaches require modification to solve difficulties such as class imbalance, noise, time complexity. Customers leaving a bank to go to a competitor is known as churn. Customers who can be predicted in advance to leave provide a firm an edge in client retention and growth. Banks may use machine learning to predict the behavior of trusted customers by assessing past data. To retain the trust of those clients, they may also introduce several unique deals. This study employed agglomerative hierarchical clustering, Decision Trees, and Random Forest Classification techniques. The data with decision tree obtained an accuracy of 84%, the data with the Random Forest obtained an accuracy of 85% and the clustered data passed through the agglomerative hierarchical clustering obtained an accuracy of 98.3% using random forest classifier and an accuracy of 98.1 % using decision tree classifier.