A SMART MARKETPLACE WITH AI-DRIVEN FRAUD DETECTION FOR FARMER-TO-BUYER TRANSACTIONS

The increasing prevalence of online transactions has heightened the risk of fraudulent activities, posing significant threats to financial security and consumer trust. Fraud detection refers to the process of identifying and preventing unauthorized or deceptive transactions, while E-payment transactions encompass digital financial exchanges conducted over electronic platforms. Traditional fraud detection methods often fail to address the evolving sophistication of fraudulent techniques, leading to substantial financial losses. This research aims to develop an advanced fraud detection system for secure E-payment transactions by leveraging machine learning algorithms computational models that learn patterns from data to improve decision-making. The study focuses on analyzing transaction patterns, identifying anomalies, and mitigating false positives (incorrect fraud alerts) while ensuring seamless user experience. A hybrid approach combining supervised learning (training models on labeled fraud data) and unsupervised learning (detecting hidden patterns without predefined labels) was employed, utilizing datasets from real-world e-payment transactions for model training and validation. Experimental results demonstrated a high detection rate with minimal false alarms, significantly outperforming conventional rule-based systems (static fraud detection using predefined rules). The study recommends integrating adaptive learning (continuously updating models with new data) to address emerging fraud trends, providing a robust solution for secure digital payments.