Machine learning is a strategy that enable computers to automatize information-driven model building and programming through a scientific discovery of statistically important patterns within the obtainable data. The learning capability of a machine and the ability to do predictive analysis is very obligatory in this age of vast information. In this study, we focused on banking sector where too many individuals are applying for bank credits. Though, it is really troublesome to determine whom loan should be granted or whom should be rejected. For banking organizations acceptance of loan is a main task. The prediction model that we formed in this paper for predicting fraudulent loan requests. In this paper, we were working with six algorithms - Decision tree, Support vector machine, Random forest, K nearest neighbors, Ada-Boost, and Logistic regression to predict the fraudulent loan request from customers. We got 83.75% accuracy from K-Nearest Neighbors algorithm which was better than other five machine learning approaches.