Analysis Of Possible Influencing Factors and Forecasting the Concentration of PM2.5

Abstract

Air quality is an important factor that affects the quality of people’s lives. With the development of the times, people’s health awareness is increasing. To truly enhance the well-being of all mankind, the author wants to figure out possible influencing factors of PM2.5 and forecast the concentration of it. The author gathers data from 2015 to 2023 and uses 4 different ways in analyzing influencing factors, then the author applies ARIMA model and LSTM model in forecasting. The author finds out that O3, precipitation, average 2-minute wind speed, and average relative humidity are nearly irrelevant to PM2.5 concentration; average air pressure and average temperature have no certainty effect on PM2.5 concentration; PM10, SO2, NO2, CO have a greater impact on PM2.5 concentration, and they are all positively correlated. As for forecasting models, LSTM model is better when performing long-step forecasts while ARIMA model is more accurate when performing short-step forecasts. These conclusions can be used by governments when forecasting relative parameters and they can take more accurate actions in dealing with PM2.5 pollution with the given influencing factors and their magnitudes.