This repository contains a capstone project focused on fraud detection in the banking sector using PostgreSQL. By querying transactional data from a fictional financial institution—PioneerTrust Bank—the project uncovers suspicious patterns and provides actionable fraud prevention recommendations.
Business Case
PioneerTrust Bank was facing an increasing volume of flagged transactions resulting in financial loss and reduced customer trust. This project uses SQL to analyze transaction patterns and support real-time fraud detection.
Tools Used
- PostgreSQL
- SQL
High-Value Transactions
- Top 5 highest transactions (₦15,000)
- Average transaction amount (₦6,650.00)
- Outliers exceeding average flagged for review
Repeat Offenders
- Identified customers with multiple flagged transactions
- Notable: A user with 5 fraud alerts
- Fraud spikes on March 2nd
Location-Based Anomalies
- Customers transacting from 3+ locations in one day
- High-risk locations: San Francisco & New York
Transaction Type Analysis
- Flagged deposits exceeded withdrawals
- Indicates vulnerabilities in deposit systems
Account Risk Profiling
- Individual accounts flagged more than business accounts
- Recommendation: tighten KYC for individual accounts
- Behavior-based fraud detection triggers
- Automated alerts for high-risk transactions
- Geo-location alerts and time-based fraud insights
- Improved account monitoring policies
PioneerTrust_db case study.sql– Core SQL queries used for analysisSQL question screenshots/– Screenshots of pgAdmin4 results and outputsSQL_Fraud_Detection_Capstone_Presentation.pdf– Summary of insights and management recommendationsREADME.md– Project overview and documentation
SQL & Data Analytics Enthusiast | Fraud Risk Analyst
📫 [[email protected]]
🔗 LinkedIn Profile
This project is open for learning and educational purposes. Attribution is appreciated.