Predictive Fraud Prevention in UPI Payment Ecosystems Using Random forest

The rapid growth of Unified Payments Interface (UPI)‑based digital transactions has transformed modern financial ecosystems, but it has also expanded opportunities for sophisticated financial fraud. A key systemic vulnerability lies in the lack of robust recipient validation prior to fund transfer, which enables fraudsters to exploit social engineering and phishing attacks. This work proposes a novel, automated fraud‑prevention framework that shifts the security focus from traditional sender authentication to receiver‑centric risk assessment. The model combines multidimensional features such as historical transaction velocity, account longevity, and aggregated risk profiles derived from reported grievance records into a dynamic fraud‑score mechanism. Experimental evaluation on synthetic UPI‑like transaction datasets demonstrates high classification accuracy in identifying high‑risk transfers, indicating strong potential for real‑time deployment in digital‑payment systems.The accuracy of the project is 95%.