Air Quality Analysis And Prediction Based On Ensemble Learning And Time Series

In recent years,due to the rapid development of industrialization,air pollution has become more and more prominent.Haze,global warming and photochemistry in the air have become key issues of global concern.In order to better prevent and treat the existing problem of air pollution,there was a distinct improvement in air quality,which requires people to know about air quality better,need to study the causes of air pollution and the change trend of air quality,and the air quality data through modern technology reasonable research analysis and forecasting.In this paper,the air quality data of Zhengzhou in recent years were analyzed and the air quality of Zhengzhou was predicted based on the technology of big data.The main research contents are as follows.(1)Statistical analysis and visualization of Zhengzhou's air quality data.General statistical analysis of air quality data sets;Visualization and analysis of daily,monthly,quarterly and year-by-year changes of the time variation characteristics of air quality index;Analysis of the proportion of five years of data for each grade of air quality;The correlation between air quality index and main air pollutants are briefly analyzed.Through research and analysis,the characteristics of the temporal variation of air quality and the important factors affecting the air quality index are obtained.(2)Integrated learning model prediction of air quality index.Several basic learning tools were used to model and predict the air quality index,e.g.Bagging method and random forest method which used to model and predict the air quality index.The prediction results of the above methods were compared and analyzed by several criteria of model evaluation,and the effective conclusion was obtained.(3)Time series model prediction of air quality index.The ARMA model and LSTM model were used to model and predict the air quality index based on time series,and the prediction results of the two models were compared and analyzed by the model evaluation criteria.(4)According to the prediction results of the integrated learning model and the time series model on the air quality index,the model evaluation criteria were used for comparative analysis.Finally,the conclusion was drawn according to the research.