Your motor insurance quote, in under a minute
Real-time motor insurance pricing powered by frequency–severity machine learning models deployed on AWS Lambda and API Gateway. This demonstration system mirrors production-style pricing architecture used in regulated insurance markets.
Try live pricing demoWhy this project matters
This demo shows how real-world motor insurance pricing can be implemented as a production-style ML system.
- Real-time motor insurance quote generation from a browser UI
- Frequency–severity pricing approach (Negative Binomial + Gamma GLM)
- Serverless deployment on AWS Lambda + API Gateway
Quote request
All fields are required. Your quote is computed in real time by the pricing engine.
Your quote
Main rating factors
- Driver age
- Vehicle type and power
- Location (region & area)
- Population density
- Bonus–Malus (no-claims level)
Model governance
- Training dataset: freMTPL2 (frequency & severity tables)
- Frequency model: Negative Binomial GLM
- Severity model: Gamma GLM (log link)
- Model version: see quote metadata above
Why this quote?
Demonstration system built to showcase production-style motor pricing architecture, not a commercial insurance offer.
We couldn’t get a quote right now.
Make sure the pricing API is running (see web-demo/README.md).
How it works
End-to-end flow from your input to the quote.
Tech stack
Backend
Python, FastAPI, AWS Lambda, API Gateway
Models
statsmodels (NB GLM, Gamma GLM), patsy, scipy
Deployment
Docker, AWS Lambda, API Gateway
Frontend
HTML, CSS, JavaScript (vanilla)
Engineering highlights
- Versioned model artifacts and pricing config
- Structured Lambda response with audit metadata
- Real-time inference via API Gateway
- Containerized deployment with Docker
- Explainable pricing (GLM term contributions)
About this project
What this demonstrates
- Frequency–severity pricing workflow
- Pure premium and gross premium calculation
- AWS Lambda + API Gateway deployment
- Explainable pricing outputs
Business relevance
- Production-style insurance pricing demo
- Real-time quote delivery
- Risk-based premium calculation
- Transparent pricing breakdown
This page is part of a portfolio project. The engine uses the freMTPL2 dataset (France) and is for demonstration only — no contract is formed.
Explore repository on GitHubBuilt by
Iman Badrooh — Applied Data Scientist / ML Engineer focused on pricing analytics and production ML systems.
- Insurance pricing and frequency–severity modelling
- End-to-end ML deployment on AWS Lambda + API Gateway
- Production-oriented data and model governance