Insurance Risk Analytics

This repository contains code and analysis for the 10 Academy Week 1 Challenge, focusing on insurance risk analytics using Python, with exploratory data analysis (EDA), data versioning, hypothesis testing, and predictive modeling.

Setup

1. Clone the repo: git clone https://github.com/Bisrath1/Insurance-risk-analytics.git
2. Create virtual environment: python3 -m venv venv
3. Activate:
   • On Unix/Linux/MacOS: source venv/bin/activate
   • On Windows: venv\Scripts\activate
4. Install dependencies: pip install -r requirements.txt

Structure

• .dvc/: Configuration for DVC local remote storage.
• SM/data/: Synthetic claims dataset managed with DVC.
• data/: Insurance datasets (insurance.csv, insurance_v2.csv) versioned with DVC.
• figures/: Visualizations from modeling and analysis (e.g., SHAP plots).
• notebooks/: Jupyter notebooks for EDA and analysis.
• results/: Outputs from regression and classification models with SHAP analysis.
• src/: Python scripts for reusable code (e.g., modeling, data processing).
• task_3_hypothesis_testing/: Scripts and results for A/B testing.
• .dvcignore: Files/folders excluded from DVC tracking.
• LICENSE: MIT License for the project.
• requirements.txt: Python dependencies, including DVC.

Progress

• Task 1: Git and EDA
  • Set up repository: https://github.com/Bisrath1/Insurance-risk-analytics.
  • Performed EDA on insurance.csv, analyzing charges by region, sex, and smoker status (branch: task-1).
• Task 2: DVC
  • Initialized DVC for data version control.
  • Versioned insurance.csv and insurance_v2.csv using DVC.
  • Configured local DVC remote storage.
  • Added synthetic claims dataset to SM/data/ (commit: 3 weeks ago).
• Task 3: A/B Testing
  • Conducted hypothesis testing on insurance charges by region and sex (branch: task-3).
  • Findings: Significant differences in charges observed in the southeast region (details in task_3_hypothesis_testing/).
• Task 4: Modeling
  • Built regression and classification models for claim severity, premium, and claim probability (branch: task-4, merged in commit: 3 weeks ago).
  • Best model: XGBoost (RMSE: X, R²: Y).
  • SHAP analysis: Smoker status and age identified as most influential features (visualizations in figures/).

Usage

• Run Jupyter notebooks for EDA and visualizations:

  jupyter lab

• Fetch versioned data with DVC:

  dvc pull

• Execute modeling scripts:

  python src/models.py

• Run hypothesis tests:

  python task_3_hypothesis_testing/run_tests.py

Recommendations

• Adjust insurance premiums for smokers and high-risk regions based on EDA and modeling insights.
• Implement claim probability model for risk-based pricing strategies.

Contributing

1. Fork the repository.
2. Create a feature branch: git checkout -b feature-branch
3. Commit changes: git commit -m "Add feature"
4. Push to the branch: git push origin feature-branch
5. Open a pull request.

License

This project is licensed under the MIT License.

Contact

Maintained by Bisrath1. For questions or feedback, open a GitHub issue or contact the owner directly.