Belige Madhumitha, Jakkidi Bhavya Sri, Nagapuri Pranay, Janne Charan, Mr. G L Singh 03 05 With the rapid growth of digital payments and online banking detecting fraudulent transactions has become an important concern for financial organizations Traditional detection methods often fail to identify fraud accurately due to the complex and evolving nature of fraudulent activities This project presents an improved fraud detection system using the XGBoost machine learning algorithm combined with Principal Component Analysis PCA to enhance feature selection and reduce data complexity The model is developed and tested using a publicly available fraud transaction dataset from Kaggle During implementation various stages such as data preprocessing feature extraction model training and evaluation are carried out systematically The performance of the proposed model is assessed using key evaluation metrics including accuracy precision recall and F1score To validate its effectiveness the results are compared with the Nave Bayes algorithm The comparison shows that the XGBoost model performs significantly better in identifying fraudulent transactions achieving accuracy above 99 along with improved precision and recall values Additionally the system is capable of predicting fraudulent activities in realtime using unseen test data Overall the results demonstrate that the proposed approach is efficient and reliable for detecting online transaction fraud making it suitable for realworld financial applications Copyright (c) Journal of Science Engineering Technology and Management Science. All rights reserved 2026