End-to-end credit risk analytics project analysing bank loan portfolios using classification techniques, statistical analysis, and automated KPI reporting. Built to simulate the kind of risk assessment workflows used in financial services and fintech analytics teams.
| Metric | Outcome |
|---|---|
| Portfolio coverage | Full loan dataset analysed across borrower segments and loan types |
| Risk classification | Loans segmented into risk tiers using classification techniques |
| Delinquency insight | Trend analysis identifying key predictors of loan default |
| Reporting automation | KPI calculations fully automated — eliminating manual spreadsheet work |
| Stakeholder output | Summary statistics and visualisations produced for non-technical audience |
Financial-Risk-Assessment-Loan-Analytics/
│
├── loan_risk_analysis.ipynb # Main notebook: EDA, classification, KPI reporting
├── data/
│ └── loan_data.csv # Source dataset (or Kaggle download link below)
├── charts/
│ ├── delinquency_trends.png # Exported visualisations
│ └── risk_tier_distribution.png
└── README.md # Project documentation
Source: Kaggle — Bank Loan Status Dataset
Key fields:
| Column | Description |
|---|---|
loan_status |
Fully Paid / Charged Off / Default |
loan_amnt |
Requested loan amount |
int_rate |
Interest rate assigned |
annual_inc |
Borrower annual income |
dti |
Debt-to-income ratio |
grade |
Risk grade assigned by lender (A–G) |
purpose |
Loan purpose (debt consolidation, home improvement, etc.) |
- Distribution of loan amounts, interest rates, and income levels
- Missing value treatment and outlier detection
- Correlation heatmap of key financial variables
- Target variable:
loan_status(binary: Paid vs. Default) - Features engineered: DTI ratio bands, income tiers, loan grade encoding
- Models evaluated: Logistic Regression, Decision Tree, Random Forest
- Best model accuracy: ~85% on test set
- Time-series view of defaults by quarter
- Risk grade (A–G) breakdown — showing default rate increases sharply below Grade C
- Loan purpose analysis — identifying highest-risk borrowing categories
- Default rate by loan grade, term, and purpose
- Portfolio health summary: total exposure, at-risk amount, recovery estimate
- Output formatted for non-technical stakeholder consumption
Add your chart PNGs here once exported from the notebook.
# Export charts from notebook with:
plt.savefig('charts/delinquency_trends.png', dpi=150, bbox_inches='tight')
# 1. Clone the repo
git clone https://github.com/rajat9526/Financial-Risk-Assessment-Loan-Analytics.git
cd Financial-Risk-Assessment-Loan-Analytics
# 2. Install dependencies
pip install pandas numpy matplotlib seaborn scikit-learn jupyter
# 3. Download dataset from Kaggle and place in data/ folder
# 4. Open notebook
jupyter notebook loan_risk_analysis.ipynb- Loans graded D and below have a default rate 3x higher than Grade A loans
- Debt consolidation loans represent the largest category — but also carry disproportionate default risk
- Borrowers with DTI > 20% show significantly higher delinquency rates
- Interest rate is the single strongest predictor of default likelihood in the dataset
python credit-risk loan-analytics classification logistic-regression random-forest pandas scikit-learn jupyter-notebook financial-analytics eda risk-assessment
Rajat Saini — Data Analyst | MBA Business Analytics LinkedIn · Portfolio · Email