NabloP/b5w3-insurance-risk-modelling-challenge

Week 3 submission for the 10 Academy Insurance Risk Analytics Challenge. Implements a modular pipeline for exploring, testing, and modeling South African car insurance claims. Features include advanced EDA, statistical A/B testing, reproducible pipelines with DVC, and machine learning models for premium optimization and claim severity prediction. # B5W3: Insurance Risk Analytics & Predictive Modeling Week 3 - 10 Academy ## 🗂 Challenge Context This repository documents the submission for 10 Academy’s **B5W3: Insurance Risk Analytics & Predictive Modeling** challenge. The goal is to support AlphaCare Insurance Solutions (ACIS) in optimizing underwriting and pricing by analyzing customer, vehicle, and claims data to: - Identify low-risk customer segments - Predict future risk exposure - Enable data-driven premium optimization This project simulates the role of a risk analyst at AlphaCare Insurance Solutions (ACIS), supporting actuarial and underwriting teams with data-driven insights for optimizing premium pricing and minimizing claims exposure. The project includes: - 🧹 Clean and structured ingestion of raw customer, vehicle, and claims datasets - 📊 Multi-layered Exploratory Data Analysis (EDA) across customer, product, geographic, and vehicle dimensions - 🧠 Modular profiling of loss ratio, outliers, and segment-specific profitability - 🗃️ Defensive schema auditing and data quality validation routines - 📦 Reproducible data versioning using DVC with Git and local cache integration - 🧪 Scaffolded modeling pipeline for classification-based claims risk prediction (planned) - ✅ Structured orchestration of insights through testable, class-based Python modules and `eda_orchestrator.py` runner script ## 🔧 Project Setup To reproduce this environment: 1. Clone the repository: ```bash git clone github.com cd b5w3-insurance-risk-modelling-challenge ``` 2. Create and activate a virtual environment: **On Windows:** ```bash python -m venv insurance-challenge .\insurance-challenge\Scripts\Activate.ps1 ``` **On macOS/Linux:** ```bash python3 -m venv insurance-challenge source insurance-challenge/bin/activate ``` 3. Install dependencies ```bash pip install -r requirements.txt ``` ## ⚙️ CI/CD (GitHub Actions) This project uses GitHub Actions fo …