Gaussian Process Regression - Pollution Prediction

This project explored using Gaussian Process Regression to model air quality in a given area based on noisy real-world data, and predict the concentration of pollutants at unmeasured locations in an urban area. Our proposed solution addressed several issues associated with Gaussian process regression, including scaling/computational cost and hyperparameter selection. The solution was selected as the best solution among 297 teams, in the course Probabilistic AI taught by Prof. Andreas Krause, Fall 2024, ETH Zürich.