Chimney Fire Risk Prediction Application

Overview

This project presents a data-driven application designed to predict chimney fire risks in the Twente region. Leveraging the methodologies from Lu et al., the application utilizes spatio-temporal modeling to assess fire risks based on variables such as building characteristics and weather conditions.

Key Features

• Risk Assessment: Evaluates chimney fire risk across different regions and timeframes.
• Weather Data Integration: Incorporates real-time weather forecasts to enhance prediction accuracy.
• Uncertainty Quantification: Provides confidence intervals alongside risk predictions.
• Interactive Visualization: Offers an interactive map interface for visualizing risk assessments.
• Model Refitting: Allows users to update model coefficients based on the latest incident, building and weather data.

Technology Stack

• Backend: Node.js with Express, R for statistical computations, Docker for containerization.
• Frontend: React with Leaflet for mapping, Tanstack Query for data fetching, Tailwind CSS for styling.
• Deployment: Hosted on Microsoft Azure using Docker containers for integration with the BI tools of Twente Fire Brigade.

Screenshots

Prediction Map View

Displays predicted chimney fire risks across different regions in the Twente area.

Model Refitting Interface

Allows the user to retrain the predictive mathematical model in four steps.

License

This project is proprietary and is protected under copyright.