JP Morgan Data Science & Analytics Projects

This workspace contains a suite of notebooks for natural gas price modeling, storage contract valuation, and loan default risk analytics. Each notebook addresses a distinct financial modeling challenge using real-world datasets and machine learning.

Project Structure

• jp_morgan.ipynb / jp_morgan_2.ipynb
  Natural Gas Price Forecasting & Storage Contract Valuation

  • Loads and processes historical natural gas prices from Nat_Gas.csv.
  • Builds a Random Forest regression model to estimate future prices.
  • Simulates and values storage contracts with customizable injection/withdrawal schedules and costs.
  • Includes scenario analysis and visualizations of price trends and seasonality.

• jp_morgan_3.ipynb
  Loan Default Risk Modeling

  • Loads loan data from Task 3 and 4_Loan_Data.csv.
  • Engineers features for credit risk analysis.
  • Trains and compares multiple classification models (Logistic Regression, Random Forest, Gradient Boosting, SVM).
  • Evaluates model performance with ROC curves, confusion matrices, and precision-recall analysis.
  • Provides functions for expected loss calculation and portfolio-level risk assessment.

• jp_morgan_4.ipynb
  FICO Score Quantization & Rating Map Creation

  • Analyzes the relationship between FICO scores and default rates.
  • Implements quantization algorithms (Uniform, MSE, Log-Likelihood) to create optimal FICO rating buckets.
  • Visualizes bucket statistics and default rate discrimination.
  • Generates rating maps for use in credit risk models.

• Nat_Gas.csv
  Historical natural gas price data.

• Task 3 and 4_Loan_Data.csv
  Loan-level data for credit risk modeling.

How to Use

1. Install Dependencies

   • Python 3.8+
   • Required packages: pandas, numpy, matplotlib, seaborn, scikit-learn, scipy

   pip install pandas numpy matplotlib seaborn scikit-learn scipy
   

2. Run Notebooks

   • Open any notebook in Jupyter or VS Code.
   • Ensure the relevant CSV files are in the workspace root.
   • Execute cells to reproduce analysis, modeling, and visualizations.

3. Key Functions

   • estimate_gas_price(date) – Predicts natural gas price for a given date.
   • price_storage_contract(...) – Values a storage contract and returns detailed economics.
   • calculate_expected_loss(loan_features) – Computes expected loss for a loan scenario.
   • FICOQuantizer – Class for creating and visualizing FICO rating buckets.

Example Usage

# Natural Gas Storage Contract Valuation
results = price_storage_contract(
    injection_dates=['2024-06-15', '2024-07-15'],
    withdrawal_dates=['2024-12-15', '2025-01-15'],
    injection_volumes=[50000, 50000],
    withdrawal_volumes=[50000, 50000],
    injection_rate=60000,
    withdrawal_rate=60000,
    max_storage_capacity=100000,
    storage_cost_per_day=1000,
    injection_cost_per_mmbtu=0.10,
    withdrawal_cost_per_mmbtu=0.10
)
print(f"Net Contract Value: ${results['net_contract_value']:,.2f}")

# Loan Expected Loss Calculation
example_loan = {
    'credit_lines_outstanding': 2,
    'loan_amt_outstanding': 5000,
    'total_debt_outstanding': 15000,
    'income': 60000,
    'years_employed': 3,
    'fico_score': 650,
    'debt_to_income': 15000/60000,
    'loan_to_income': 5000/60000,
    'credit_utilization': 15000/3,
    'income_per_year': 60000/3
}
result = calculate_expected_loss(example_loan)
print(f"Probability of Default: {result['probability_of_default']:.2%}")
print(f"Expected Loss: ${result['expected_loss']:.2f}")

Visualization

• All notebooks include code for plotting trends, distributions, ROC curves, and bucket statistics.

License

This workspace is for educational and analytical purposes only.