Machine Learning for Fraud Detection Leveraging SAP Data: A Case Study for ML Application

• Author(s): Ashitosh Chitnis
• Paper ID: 1707493
• Page: 399-411
• Published Date: 28-02-2022
• Published In: Iconic Research And Engineering Journals
• Publisher: IRE Journals
• e-ISSN: 2456-8880
• Volume/Issue: Volume 5 Issue 8 February-2022

Abstract

Businesses encounter serious financial fraud issues which produce major money loss and reputation destruction together with regulatory penalties. Businesses currently use traditional fraud detection tools which are based on rules but they struggle to detect modern complex fraudulent actions. Financial institutions apply Machine Learning as a strong detection tool to process vast financial data which enables them to detect suspicious operational patterns linked to fraud. The research evaluates the implementation of ML-based fraud detection on BigQuery ML which is Google Cloud's machine learning platform through SAP Finance Data. The study executes a process that includes cleans up raw data while applying new fields before choosing an ideal model for training purposes and conducting performance assessments using authentic transaction records. Numerous ML algorithms and logistic regression along with decision trees and deep learning techniques get implemented while comparing their performance using accuracy and other metrics such as precision and recall and F1-score. BigQuery ML shows its capability to detect fraud through efficient data handling capabilities allowing easy processing of large datasets in real time while minimizing system resource consumption. The documented evidence shows ML-based fraud detection outperforms conventional rule-based solutions in terms of precision, speed of detection along with flexibility in application scenarios. BigQuery ML provides enterprise-level tools for model analysis alongside deployment functionalities which enable financial institutions to implement its solutions effectively when handling their large-scale data. Even so the acceptance of these solutions requires solving issues with model interpretability problems and privacy complications along with resource limitations. The research advances understanding of AI-driven financial security through evaluation of cloud-based ML solutions which boost detector systems for financial crimes. Research results reveal the necessity of improving artificial intelligence models with new capabilities that counter fraud schemes as they develop. Studies should focus on developing fraud detection mechanisms through combining explainable AI (XAI) methods with real-time anomaly systems.

Keywords

Fraud Detection, BigQuery ML, SAP Finance Data, Machine Learning, Real-Time Analytics

Citations

IRE Journals:
Ashitosh Chitnis "Machine Learning for Fraud Detection Leveraging SAP Data: A Case Study for ML Application" Iconic Research And Engineering Journals Volume 5 Issue 8 2022 Page 399-411

IEEE:
Ashitosh Chitnis "Machine Learning for Fraud Detection Leveraging SAP Data: A Case Study for ML Application" Iconic Research And Engineering Journals, 5(8)