Detecting banking frauds with analytics and machine learning

Abstract

Bank fraud is the bodily loss of a Bank or maybe the loss of very sensitive info. For detection, there are lots of machine learning algorithms which can be used. The study shows many algorithms which could be used for deciding transactions as fraud or perhaps real. The information set employed in Bank fraud Detection was utilized in the research. The SMOTE method was used for oversampling, since the dataset was incredibly imbalanced. Moreover, include choice was performed, and the set was divided into two parts, test data and instruction information. The algorithms used in this study were Logistic Regression, Multilayer Perceptron, Random Forest and Naive Bayes. The results show that every algorithm could be used with good precision for fraud detection of banking solutions. For the detection of extra constipation, the proposed model might be used.