Global Banking Fraud Detection & Risk Analysis

Executive Summary

This project analyzes over 153,000 banking transactions to uncover fraud patterns, transaction behavior, and financial risk exposure using a scalable analytics pipeline. The objective is to support proactive fraud monitoring rather than reactive controls.

Data Cleaning & Preparation (Excel)

Raw transactional data was cleaned and standardized in Excel to ensure analytical accuracy before loading into SQL.

• Removed useless columns, empty rows, and duplicates
• Trimmed whitespace and standardized text formatting
• Split customer names into first and last names
• Cleaned date and transaction amount formats
• Created amount buckets and refund tracking columns
• Standardized merchant, category, transaction type, and country names
• Prepared dataset for SQL querying and Power BI modeling

SQL Analysis Layer

SQL acted as the analytical engine, transforming cleaned data into structured KPIs and fraud insights before visualization.

• Transaction volume and monetary flow analysis
• Fraud count, fraud rate, and average risk score calculation
• Country-wise and merchant-wise exposure analysis
• Refund and reversal behavior assessment
• Merchant category fraud rate comparison

Power BI Dashboards & Visualization

SQL outputs were connected live to Power BI to build interactive dashboards for executives, risk teams, and analysts.

• Global transaction flow and geographic exposure
• Digital vs physical transaction behavior
• Merchant category and transaction type fraud risk
• Refund behavior and fraud concentration analysis

Conclusion & Business Impact

This project demonstrates how an end-to-end analytics pipeline can convert large-scale banking data into actionable fraud intelligence. The framework mirrors real-world banking environments and supports predictive, behavior-based risk monitoring.