Abstract:
Today, digitization is turning popular because of the ease and convenience of utilizing e-commerce. People are selecting on-line payments and electronic shopping due to the convenience of time, the convenience of transportation, and so forth. As a result of the high level of e-commerce usage, fraud of credit card is increasing rapidly. Credit card transactions are very common nowadays and so is the fraud related to it. One of the most common fraud interface processes is to illegally collect the cards, user data and use the collected data for tele-ordering. Once enough info is collected and made available, it becomes challenging for an individual or any company to track down such fraud records among thousands of standard transactions. The fraud detection in credit card transactions is essential with enhanced performance measures. A methodology for effectual classification of fraudulent transactions is proposed in this paper. Also, Machine Learning (ML) algorithms like Decision tree, Random Forest, Logistic Regression and KNN are applied for fraud detections in credit card dataset. Random Forest and Decision Tree methods have shown highest accuracy with adequate F-score. The fused feature selection process is required in future to identify the significant features of the data to enhance the performance of the classifier models.
