This repository contains the deliverables for our team's submission to the Quantify 2025 Energy Risk & Insurance Case Competition, hosted by the University of Waterloo. We explored CAT event loss modelling, parametric trigger analysis, and strategic feasibility for Quant Co.'s expansion into renewable energy.
Quant Co. seeks to expand into renewable energy (solar & wind) but faces increasing CAT-related risks such as wildfires, hail and winterstorms. Our task was to assess these risks and design a parametric insurance strategy to support their expansion.
- Machine Learning classification of high-loss CAT events (Logistic Regression, Random Forest, XGBoost)
- Power BI dashboard with interactive risk insights
- Strategic recommendations for expansion by region & hazard type
- Slide deck from Power BI and QR code demo
| Model | AUC | Precision | Recall |
|---|---|---|---|
| Logistic Regression | 0.56 | 0.40 | 0.40 |
| Random Forest | 0.56 | 0.38 | 0.33 |
| XGBoost | 0.60 | 0.45 | 0.33 |
- SMOTE oversampling is used to handle class imbalance.
- Top features: max wind speed, rainfall, hail size, structures destroyed.
- Python: pandas, scikit-learn, imblearn, matplotlib)
- Machine Learning: XGBoost, Random Forest, Logistic Regression
- Power BI: Interactive visual dashboard with filters and KPIs
- Git & GitHub: Version control and collaboration
Feel free to connect or follow more of my work: