Cloud-Optimized AI Framework for Real-Time E-Commerce Fraud Detection

Asfand, - (2025) Cloud-Optimized AI Framework for Real-Time E-Commerce Fraud Detection. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (1MB) | Preview

Abstract

The growth in e-commerce transactions has witnessed an increase in fraudulent transactions; hence, efficient, accurate, and scalable fraud detection systems are needed. The proposed research is a fraud detector framework using supervised machine learning, developed using the popular credit card fraud data on Kaggle. Since the degree of class imbalance in the dataset is too high, an oversampling strategy, SMOTEENN, serves to blend oversampling with data cleaning. A classification model is trained, based on XGBoost and optimized for logloss, with a random state set to a fixed value to ensure reproducibility. Probabilistic predictions are produced and binarized to binary predictions at an optimized threshold, which is arbitrarily set at 0.4, to enhance sensitivity to minority class predictions. The model offers solid predictive strength as it acquired an accuracy of 99.97%, precision of 1.0 and 0.9993 of non-fraud and fraud transactions, respectively, and recall of 0.9993 and 1,0. The F1 scores are high in both classes and fail to show any unbalanced classification. The confusion matrix is reported as 38 false positives and 0 false negatives; thus, the performance of not misclassifying the cases, particularly the fraudulent ones, is excellent. The general architecture is designed to be deployed in various cloud services allowing the creation of a method of detection of fraud applicable to modern e-commerce systems.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Sahni, Vikas UNSPECIFIED
Uncontrolled Keywords:	Fraud Detection; E-Commerce; XGBoost; SMOTEENN; Cloud Deployment; Imbalanced Data; Real-Time Classification
Subjects:	Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence T Technology > T Technology (General) > Information Technology > Cloud computing Q Science > QA Mathematics > Computer software > Computer Security T Technology > T Technology (General) > Information Technology > Computer software > Computer Security H Social Sciences > HF Commerce > Electronic Commerce
Divisions:	School of Computing > Master of Science in Cloud Computing
Depositing User:	Ciara O'Brien
Date Deposited:	20 Mar 2026 10:16
Last Modified:	20 Mar 2026 10:16
URI:	https://norma.ncirl.ie/id/eprint/9196

Actions (login required)

View Item