Study of air quality due to Madrid Central with Gaussian Processes

Abstract

This master thesis examines the performance of Gaussian process regression models on air pollution NO2 levels six years before and six years after the implementation of the Madrid Central. The theoretical background is reviewed, including the multivariate Gaussian distribution, stochastic Gaussian processes, and covariance functions. A methodology is presented to analyze the impact of a Low Emission Zone using meteorological data

from the Retiro Park, Plaza del Carmen, and Cuatro Vientos stations. Different models

are evaluated based on the predictive ability using RMSE. The methodology classifies an

event as relevant if the RMSE values increase from one year to another.