Application Of Statistical Method In Air Quality Analysis And Prediction

At present,the situation of air pollution in China is increasingly severe,and haze weather is increasing in the north.The main characteristics of regional environmental problems are fine particles and inhalable particles,which threaten the health of human beings and the harmonious development of society.As China is in the process of transforming to industrialization and urbanization,the increasing demand for energy and resources has also raised pressure on the prevention and control of air pollution.Therefore,it is the most effective way to improve the air quality to understand the current situation of air quality in China and to analyze and explain the air quality and to take effective measures in time to improve the air quality.Air quality evaluation and air environment prediction have always been the focus of environmental management.Time series analysis is a fitting model of curve fitting and parameter estimation by combining the characteristics of data.It can not only reveal the inherent law of development of things through data,but also dynamically reveal the principles of various behaviors.The support vector machine(SVM)is to obtain a model with good generalization ability,and seeks a compromise between the complexity of the model and the learning ability according to the limited information in the data sample.Both of the two are from different angles to excavate the inherent law from the information provided by the data,and the practicality and universality are recognized in various fields,and can also be used as an important method of air quality analysis and prediction.This paper mainly uses time series analysis and support vector regression as the main analysis method.Using different criteria of model order,on average January 2014-2017 year in August in Shenyang city AQI were built in different application time series model,support vector machine to establish the knowledge system,the R software support vector regression model,grey theory and random forest regression model,also establishes some combination of several models.And in September 2017 to December 2017 in the monthly average AQI of recursive prediction,using the actual data and the forecast is the average relative error,prediction ability of model to measure the mean monthly AQI of Liaoning Province based on the principle of the minimum average error.