Time series-based PM2.5 concentration prediction in Jing-Jin-Ji area using machine learning algorithm models

• DOI: 10.1016/j.heliyon.2022.e10691
• Publication Date: 2022-09-23T08:56:19Z
• Authors: Xin Ma; Tengfei Chen; Rubing Ge; Caocao Cui; Fan Xu; Qi Lv
• Keywords: H1-99; Science (General); K-Nearest Neighbor; PM2.5 prediction; Lasso regression; Jing-Jin-Ji city group; 01 natural sciences; Linear SVR; Social sciences (General); Q1-390; Gradient boosting; Research Article; 0105 earth and related environmental sciences

Globally all countries encounter air pollution problems along their development path. As a significant indicator of quality, PM2.5 concentration has long been proven to be affecting the population's death rate. Machine learning algorithms outperform traditional statistical approaches are widely used in prediction. However research on model selection discussion and environmental interpretation prediction results is still scarce urgently needed lead policy making control. Our compared four types machine algorisms LinearSVR, K-Nearest Neighbor, Lasso regression, Gradient boosting by looking into performance predicting concentrations among different cities seasons. The show that able forecast next day based previous five days' data with better accuracy. comparative experiments city level Boosting mean absolute error (MAE) 9 ug/m3 root square (RMSE) 10.25–16.76 ug/m3, lower other three models, season models have best performances winter time worst summer time. And more importantly demonstration models' each great significance implications.