Machine learning (ML), a branch of artificial intelligence (AI), enables prediction and decision-making by learning from data. This technique forecasts complex problems by relating climatic variables to indoor pollutants, allowing reliable predictions. The study takes place in social housing in southeastern Mexico, aiming to predict indoor air quality using climatic data. A mathematical model will provide a representative function. The main goal is to develop smart sensors that require fewer pollutant sensors, reducing economic costs and improving social well-being through efficient, data-driven air quality monitoring.
A comprehensive methodology has been implemented that integrates advanced techniques, such as correlation analysis, machine learning models, and sensitivity analysis, with the aim of examining the correlation between indoor pollutants and climate variables, both indoor and outdoor. This approach enables determination, prediction, and quantification of pollutants inside the home based on climate variables.
The experimental results revealed poor indoor air quality in the dwelling under study, exceeding the regulatory limits established by the World Health Organization. Despite observing a low correlation between climate variables and pollutants, the results obtained through machine learning techniques indicate that the Ensemble trees and Gaussian process regression algorithms yielded the best results, demonstrating remarkable predictive power. Finally, the SOBOL sensitivity analysis made it possible to determine the climatic variables with the greatest impact on pollutants. The results obtained in this study will serve as a basis for the future development of smart sensors that will be capable of controlling passive and active components, with the aim of improving indoor air quality in homes.
The objective of this study is to identify and quantify the amount of pollutants inside a home. To do this, it uses climatological variables that, together with machine learning techniques, allow the type of pollutant and its concentration level inside the home to be predicted. The results of this study provide a valuable framework for the development of smart sensors.
