Downloads: 0
Analysis Study Research Paper | Information Technology | Volume 15 Issue 4, April 2026 | Pages: 1441 - 1444 | India
A Predictive Model for Detecting Fraudulent Transactions in Financial Systems
Abstract: Global financial security is seriously threatened by the rise in fraudulent activities brought on by the development of digital financial systems. Conventional rule-based systems are unable to adjust to changing fraud trends. The Light Gradient Boosting Machine (LightGBM) algorithm is used in this paper's machine learning-based financial fraud detection system, which is coupled with a Streamlit dashboard for real-time prediction and visualization. The system employs AutoML for hyperparameter optimization and SMOTE for data balancing. The trained LightGBM model accurately and interpretably classifies transactions as either fraudulent or legitimate. The interactive Streamlit interface offers analytical insights and visualizes fraud trends. This model is appropriate for real-world financial fraud detection because it exhibits scalability, transparency, and real-time responsiveness.
Keywords: Fraud Detection, LightGBM, Machine Learning, Streamlit, SMOTE, AutoML, Digital Security
How to Cite?: S. A. Bagul, Vaibhavi Handibag, Yash Lawande, Prem Mandhare, Shrikant Nevse, "A Predictive Model for Detecting Fraudulent Transactions in Financial Systems", Volume 15 Issue 4, April 2026, International Journal of Science and Research (IJSR), Pages: 1441-1444, https://www.ijsr.net/getabstract.php?paperid=SR251114101640, DOI: https://dx.dx.doi.org/10.21275/SR251114101640