Authors:
M. A. Thinesh, S. S. Mukhil Varmann, S. Leoni Sharmila, Sonjoy Ranjon Das

Addresses:
1,2,Department of Computer Science and Engineering, SRM Institute of Science and Technology, Ramapuram, Chennai, Tamil Nadu, India. 3Department of Mathematics, Saveetha School of Engineering, Saveetha Institute of Medical and Technical Sciences, Thandalam, Chennai, Tamil Nadu, India. 4Department of Computing, Shipley College, Shipley, England, United Kingdom. tm9045@srmist.edu.in1, sm7225@srmist.edu.in2, leonisharmilas.sse@saveetha.com3, sanjoy.das@shipley.ac.uk4

Abstract:

Credit card fraud significantly threatens financial institutions and consumers worldwide. To address this issue, this project leverages machine learning techniques, specifically a RandomForest Classifier, to detect fraudulent credit card transactions. The dataset is from Kaggle and contains transaction details, including transaction amounts and class labels indicating fraud or non-fraudulent transactions. The project begins with data exploration and visualization to gain insights into the dataset’s characteristics. It uses various data visualization techniques, such as classification plots and correlation matrices, to understand the understood patterns. After prеprocеssing the data and dividing it into training and test sets, the random forest classifier is trained on training data. Learning curves visualize the model’s performance as the training dataset size varies. A comprehensive set of metrics is utilized to evaluate the model’s effectiveness. It includes accuracy, specificity, error rate, and a confusion matrix to assess the model’s ability to classify fraudulent and non-fraudulent transactions. In addition, precision, recall, and F1-scorе are computed. Receiver Operating Characteristic (ROC) and Prеcision-Rеcall curves are generated to give a detailed understanding of the model’s performance and to assess the power mean to discriminate between classes’ precision-recall trade-offs. The project concludes with an evaluation of the model’s performance, highlighting its strengths and areas for improvement. This project serves as a valuable example of the application of machine learning for fraud detection in financial transactions, benefiting financial institutions and consumers by reducing financial losses due to fraud and increasing security.

Keywords: Random Forest Classification Model; Accuracy and Precision; Receiver Operating Characteristic; F1-Score and Recall; Machine Learning; Confusion Matrix; Credit Card and Personal Information; Reducing Financial Losses.

Received on: 15/06/2023, Revised on: 10/09/2023, Accepted on: 11/10/2023, Published on: 24/12/2023

FMDB Transactions on Sustainable Technoprise Letters, 2023 Vol. 1 No. 4, Pages: 181-194

• Views : 147
• Downloads : 9

Download PDF