AI-Driven Indoor Temperature Prediction for Tanzanian Classrooms

Background

In Tanzania’s hot climate, classrooms often experience extreme indoor temperatures, which compromise student learning and safety. The lack of detailed data on factors such as roofing materials and ceiling boards further exacerbates the issue, as traditional methods fail to provide actionable insights for improving classroom environments. This challenge affects not only educational outcomes but also resource allocation and infrastructure policy effectiveness.

Objective

To develop an AI-driven predictive model leveraging satellite imagery and environmental data to estimate indoor classroom temperatures. This project aims to empower Tanzanian educational authorities with precise, actionable data to optimize resource allocation, infrastructure planning, and policy implementation for safer, more conducive learning environments.

Approach

The 8-week project was undertaken by Omdena team of 50 AI engineers, tackled the problem using the following key methodologies:

1. Data Collection and Integration: Gather high-resolution satellite imagery from platforms like Google Maps and Sentinel 2, alongside environmental data such as historical weather, greenery coverage, and temperature records.
2. Problem Definition and Model Development: Enhance school location data accuracy and develop algorithms to infer indoor temperatures based on external observable factors like roof type and local climate conditions.
3. Temperature Range Detection: Identify and classify classrooms into temperature ranges to highlight potential health and learning risks.
4. Visualization and Reporting: Create a geospatial dashboard with color-coded temperature profiles, aiding in quick identification of at-risk schools.
5. Additional Feature – Flood Risk Assessment: Integrate flood risk data to address broader environmental safety concerns.

Tools like advanced geospatial analysis, machine learning algorithms, and user-friendly visualization platforms ensured a comprehensive and scalable solution.

Results and Impact

• Enhanced Insights: The model accurately estimated indoor temperatures based on structural and environmental features, providing actionable insights for over 100 schools.
• Better Resource Allocation: Educational authorities can now target interventions to the most affected schools, optimizing resource use.
• Improved Learning Environments: Safer, more conducive classrooms promote better focus, comfort, and overall academic performance among students.
• Broader Policy Support: The data-driven approach supports tailored infrastructure policies, aiding long-term improvements in educational settings.
• Scalable Solution: The integration of satellite imagery ensures this model can be adapted to other regions facing similar challenges.

Future Implications

This project demonstrates the transformative potential of AI in addressing infrastructure challenges in education. Future research could integrate more granular classroom data, refine flood and extreme weather risk assessments, and expand the model’s use to other regions. Additionally, these findings could inform broader climate change adaptation strategies and inspire policies that prioritize sustainable and resilient school infrastructure.