A comprehensive multi-dimensional analysis of a banking dataset covering loan portfolios, credit-debit transaction flows, branch performance, risk exposure, and fraud indicators.
This project analyzes banking operations across 66,000+ loan accounts and millions of transactions to uncover risk patterns, customer behavior, and operational inefficiencies. The analysis spans SQL querying, Excel modeling, Power BI dashboards, and Tableau visualizations.
- Balanced Transaction Flows — Credit (₹127.60M) and debit (₹127.29M) show near-perfect equilibrium (ratio: 1.00), indicating financial stability
- Loan Portfolio — Total loan value of $751M across 66K loans; collections at $809M with a low default rate of ~1.56%
- Age Group 26–35 dominates — 50%+ of borrowers; 36-month maturity loans preferred by 95.7% of customers
- Top Performing Cities — Mathura (₹31M), Sangrur (₹25M), Agra (₹21M)
- Geographic Concentration — Heavy exposure in Uttar Pradesh → diversification risk identified
- Verification vs Default — Unverified accounts (16,548) drive the majority of defaults; KYC gaps are a critical risk driver
- High-Risk Transactions — ~20% of flagged activity identified; seasonal dips may mask fraud patterns
| Metric | Value |
|---|---|
| Payments Received | ₹482.70M |
| Loans Disbursed | ₹388.96M |
| Active Accounts | 39,717 |
| Default Rate | 2.57% |
| Total Interest | ₹89.91M |
| Tool | Usage |
|---|---|
| SQL | Data extraction, aggregations, risk segmentation queries |
| Power BI | Interactive dashboards, KPI cards, geographic maps |
| Tableau | Transaction trend analysis, branch performance visuals |
| Excel | Data cleaning, pivot tables, financial modeling |
Bank_Analytics/
├── data/ # Raw dataset (loan & transaction data)
├── sql/ # SQL queries for analysis
├── dashboards/ # Power BI (.pbix), Tableau (.twbx), Excel dashboards
└── README.md
- Risk Management — Deploy behavior-based early warning systems; recalibrate default models by customer segment
- Growth Optimization — Target 26–35 demographic and high-performing cities; adjust product pricing by branch profitability
- Collections Strategy — Use digital reminders, hardship programs, and weekly cure-rate monitoring
- Fraud/AML Prevention — Strengthen detection rules during seasonal dips; monitor suspicious transaction spikes
- Enhanced KYC — Verification status is the strongest predictor of default → stricter onboarding needed
Sarfaraz Ahmad
Data Analyst · SQL · Python · Power BI · Tableau
GitHub | LinkedIn