Statistical Analysis Of PM_2.5 Data

The air deterioration and the frequently apparence of haze frequently attracted sus-tained attention to PM2.5. This article from the perspective of time series analysis, the PM2.5data of each city in the analysis, including smooth, randomness, heteroscedas-ticity. First we use ARMA model to fit the PM2.5sequence, and then according to the sequence of the heteroscedasticity of fitting the ARMA-GARCH model to describe the sequence character. Considering the limitation of ARMA linear function, we use the neural network model of the sequence, and the GARCH fitting of the residuals, re-ceived good results. According to the results of the analysis and received during the analysis of PM2.5sequence was inspired, according to city and by the time we got the two kinds of classification, so as to better describe only the properties of the PM2.5sequence analysis from the point of data.