Investigating Fraudulent E-Commerce Transactions: A Data-driven Approach Using Machine Learning
Joel Buehler · 2024
Detecting fraudulent activities within e-commerce transactions using data analytics and machine learning is critical for protecting businesses and consumers. Fraudulent transactions significantly threaten online platforms, leading to financial losses, inventory depletion, and a decline in consumer trust. Fraud detection systems aim to classify transactions as legitimate or fraudulent, helping companies safeguard against these risks. Machine learning algorithms can analyze patterns within transaction data, identifying anomalies that indicate potential fraud. By leveraging these technologies, businesses can implement automated, efficient, and accurate systems to detect fraud in real-time, saving resources and enhancing overall security. This research project focuses on developing a machine-learning model capable of predicting fraudulent e-commerce transactions. The project aims to distinguish between legitimate and fraudulent transactions using a dataset that includes various transactional attributes, such as purchase amounts, timestamps, user demographics, and behavioral patterns. The machine learning models employed in this study include Neural Networks, Gradient Boosting, Random Forest, Logistic Regression, and Support Vector Classification (SVC). These models are trained on a large dataset, and their performance is evaluated using accuracy, precision, recall, and F1-score metrics. This research aims to identify patterns and trends within fraudulent transactions and develop a robust model for their detection. Based on the evaluation metrics, the aim is to determine which machine learning model most effectively distinguishes fraudulent transactions from legitimate ones. The results are expected to provide valuable insights into the characteristics of fraudulent transactions and contribute to developing more sophisticated fraud detection systems for e-commerce platforms.