Air Quality Characterization Using k-Nearest Neighbors Machine Learning Algorithm via Classification and Regression Training in R
Authors: Timothy M. Amado
Abstract
Through the years, environmental health and protection have been ignored. However, because of recent phenomena such as climate change, people are slowly becoming aware of the environment. One of the main concerns nowadays is air pollution. To this avail, the U.S. Environmental Protection Agency (EPA) standardized air quality with the use of air quality index (AQI). However, AQI requires accurate sensor readings and complex calculation to obtain. Hence, the objective of this paper is to solve that problem by characterizing the air quality with regards to AQI through the use of k-nearest neighbors machine learning algorithm. The proposed methodology is implemented using a prototype of integrated gas sensors for data gathering. R programming, focusing on classification and regression training (caret) package for data processing, model development, and algorithm tuning, is utilized. The system is evaluated, and an accuracy of 99.56% is obtained.