Tools
Tools
Ahire, Pranav Dinesh (2025) Real-Time AI-Powered Anomaly Detection in Financial Transactions Using Microsoft Fabric and Power BI. Masters thesis, Dublin, National College of Ireland.
Preview |
PDF (Master of Science)
Download (2MB) | Preview |
Preview |
PDF (Configuration Manual)
Download (2MB) | Preview |
Abstract
The high influx of digital activities in transactions has contributed to a significant increase in the risk and sophistication of financial fraud that has become a real challenge to the financial institutions that apply traditional rule-based detection methods. Such legacy models usually yield too many false positives, and are not real-time responsive. This study fills these gaps with designing and deploying actual real time AI powered fraud detection solution with the help of Microsoft Fabric and Power BI that is free of any third party tools like Python or R that would be required otherwise.
The research aims at proposing a new end-to-end pipeline combining both OneLake storage and Synapse ML and AI-driven algorithms of anomaly detection using mechanisms such as Isolation Forests, Artificial Neural Networks (ANNs), and Support Vector Machines (SVMs) to detect suspicious patterns in financial transactions. Mass personalization, anomaly classification and fraud risk scoring are enabled by the system, which contains custom ETL pipelines and feature engineering deployed in Microsoft Fabric, displayed in real time in Power BI dashboards.
The tested results prove that the AI- and ML-based models can be used when implemented natively in Power BI to cut the false positives and drastically increase the accuracy and efficiency of fraud detection. It is also integrated into human in the loop decision making by which fraud analysts can retrain, refine, and review the AI models on an ongoing basis. Such integration provides a fraud monitoring system with an ethical and transparent approach in accordance with GDPR.
This piece of work plays a role in the emerging real-time fraud detection via AI augmented analytics to Business Intelligence domain, demonstrating viability, elasticity, and accuracy of detecting frauds and the necessities to harness the integration of data science and decision-making to financial intelligence.
| Item Type: | Thesis (Masters) |
|---|---|
| Supervisors: | Name Email Ain, Qurrat Ul UNSPECIFIED |
| Uncontrolled Keywords: | Financial fraud detection; Anomaly detection; Microsoft fabric; Synapse ML; Isolation forest; Artificial Neural network (ANN); Support vector machine (SVM); Power BI; Real-time monitoring; AI-human collaboration; False Positive reduction |
| Subjects: | H Social Sciences > HG Finance 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 Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning |
| Divisions: | School of Computing > Master of Science in Data Analytics |
| Depositing User: | Ciara O'Brien |
| Date Deposited: | 24 Aug 2026 14:36 |
| Last Modified: | 24 Aug 2026 14:36 |
| URI: | https://norma.ncirl.ie/id/eprint/9614 |
Actions (login required)
 |
View Item |