COMPARATIVE ANALYSIS OF MACHINE LEARNING MODELS FOR PREDICTION OF AIR POPULATION

초록

"This study analyzes the correlation between air pollutants in Korea and in Jiangsu, Hebei, and Shandong provinces in China, which are closest to Korea. The regression models in this study predict the amount of sulfur dioxide, carbon monoxide, ozone, nitrogen dioxide, particulate matter, and ultra-particulate matter in the atmosphere of the Korean peninsula. We use linear regression, k-nearest neighbor, AdaBoost, gradient boost, random forest, bagging, and XGBoost algorithms for predictive regression models. Through feature importance, we confirm that Jiangsu’s air pollutants have the most significant effect on the atmosphere of the Korean peninsula and identify the importance of other independent features. We evaluate and compare the results of six models using performance measures of R2-Score, mean squared error, root mean squared error, and mean absolute error. The model using XGBoost shows the best results. © 2023 ICIC International.

키워드