Scalable and Explainable Credit Card Fraud Detection Using ML

Abstract

Our project primarily focuses on real-world credit card fraud detection. To begin, we will collect credit card datasets to serve as the training dataset. Subsequently, we will provide user credit card queries as the testing dataset. The classification process will be carried out using the Random Forest algorithm, which will analyse both the pre-existing dataset and the newly provided user data. The ultimate goal is to optimize the accuracy of the results. Additionally, we will process specific attributes to detect fraudulent activities and present the findings using graphical model visualization. The performance of the applied techniques will be evaluated based on key metrics, including accuracy, sensitivity, specificity, and precision.

Introduction

Credit card fraud is a significant and growing threat in the digital economy, causing financial losses and undermining security. This project aims to improve fraud detection by using machine learning, specifically Logistic Regression, to classify transactions as fraudulent or legitimate based on features extracted from transaction data.

The system uses a balanced, anonymized real-world dataset to train the model, achieving high accuracy in detecting suspicious transactions. It features a user-friendly Flask-based web interface that allows users to input transaction details and receive real-time fraud predictions. The project integrates libraries like NumPy, Pandas, and Scikit-learn for data handling and modeling, and Matplotlib and Seaborn for visualization. A secure login system ensures data protection.

The main objectives include minimizing false positives and negatives to reduce financial losses and protect customers, providing real-time detection, and enabling continuous model learning to adapt to evolving fraud tactics. The methodology involves data acquisition, preprocessing (including balancing), feature extraction, classification using Logistic Regression, real-time prediction, and performance evaluation using metrics like accuracy, precision, recall, and F1-score.

The implementation covers data collection, preprocessing, feature extraction, and model evaluation using hold-out and cross-validation techniques. The project also extends previous work by integrating the Logistic Regression model with a web interface and secure user authentication. Feature extraction focuses on transaction patterns such as amount, time, and frequency to detect anomalies, and transfer learning is suggested for deep learning approaches to enhance performance with limited data.

Conclusion

This project successfully demonstrates the application of machine learning in financial security by developing an efficient Credit Card Fraud Detection system. Utilizing TensorFlow/Keras and Scikit-learn for model training, along with advanced preprocessing techniques for transaction data, the system ensures accurate, real-time fraud detection. A Flask-based web interface allows users and financial institutions to interact with the model seamlessly, while SQLite is used for secure and efficient transaction data management. The results indicate that machine learning models can significantly enhance fraud detection by identifying suspicious transactions with high precision and recall, thereby helping financial institutions reduce losses and protect customer accounts. While the current implementation delivers strong performance, future enhancements could involve incorporating real-time big data processing, deploying the model at scale with cloud infrastructure, and integrating explainable AI techniques to improve transparency and trust. This work highlights the critical role of artificial intelligence in the financial sector, showing how technology can complement existing fraud prevention systems to improve accuracy, scalabili ty, and response times in combating financial crime.

References

[1] P. Richhariya and P. K. Singh, “Evaluating and emerging payment card fraud challenges and resolution,” International Journal of Computer Applications, vol. 107, no. 14, pp. 5 – 10, 2014. [2] S. Bhattacharyya, S. Jha, K. Tharakunnel, and J. C. Westland, “Data mining for credit card fraud: A comparative study,” Decision Support Systems, vol. 50, no. 3, pp. 602–613, 2011. [3] A.DalPozzolo,O.Caelen,Y.-A.LeBorgne,S.Waterschoot,andG.Bontempi, “Learned lessons in credit card fraud detection from a practitioner perspective,” Expert systems with applications, vol. 41, no. 10, pp. 4915– 4928, 2014. [4] C. Phua, D. Alahakoon, and V. Lee, “Minority report in fraud detection: classi?cation of skewed data,” ACM SIGKDD explorations newsletter, vol. 6, no. 1, pp. 50–59, 2004. [5] Z.-H. Zhou and X.-Y. Liu, “Training cost-sensitive neural networks with methods addressing the class imbalance problem,” IEEE Transactions on Knowledge and Data Engineering, vol. 18, no. 1, pp. 63–77, 2006. [6] S. Ertekin, J. Huang, and C. L. Giles, “Active learning for class imbalance problem,” The 30th annual international ACM SIGIR conference on Research and development in information retrieval, pp. 823–824, 2007. [7] M.WasikowskiandX.-w.Chen,“Combatingthesmallsampleclassimbalance problem using feature selection,” IEEE Transactions on knowledge and data engineering, vol. 22, no. 10, pp. 1388–1400, 2010. [8] S. Wang and X. Yao, “Multiclass imbalance problems: Analysis and potential solutions,” IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics), vol. 42, no. 4, pp. 1119–1130, 2012. [9] R. J. Bolton and D. J. Hand, “Statistical fraud detection: A review,” Statistical science, pp. 235–249, 2002. [10] D. J. Weston, D. J. Hand, N. M. Adams, and C. Whitrow, “Plastic card fraud detection using peer group analysis,” vol. 2, pp. 45–62, 2008.

Copyright

Copyright © 2025 Mrs. CH Swapna, Jayavarapu Sruthi, Katamgari Tirumala, Jiddu Venkata Abhilashh, Kanuru Chechala Jai Vidya. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.