Securing the Digital Realm: Unmasking Fraud in Online Transactions Using Supervised Machine Learning Techniques

A key component of contemporary banking systems and e-commerce platforms is identifying fraud in online transactions. Traditional rule-based techniques are insufficient for preventing sophisticated fraud schemes because of the increasing complexity and number of expanding online transactions. This research study examines the development of fraud detection methods, emphasizing data analytics and machine learning (ML) models. The study also focuses on the fact that developing efficient fraud detection systems requires continuous observation, data preprocessing, feature selection, and testing of models. Seven ML models, Logistic Regression (LR), Decision Trees (DT), k-Nearest Neighbors (kNN), Naïve Bayes (NB), Support Vector Machine (SVM), Random Forests (RF), and Extreme Gradient Boosting (XGBoost) are considered for classifying the dataset into fraudulent or not. During the experimentation study, it was observed that XGBoost yielded the highest accuracy of 99% when compared to other models. Users can determine which features significantly influence the model's predictions by using XGBoost's feature significance insights. Additionally, XGBoost provides integrated support for managing missing values in data, negating the requirement for imputation and other preprocessing procedures. Due to these, it performed better.