Analysis And Evaluation Of The Air Quality Index In Shang Hai Based On Multivariat Statistics And Intelligence Algorithm

We begin with analysis of AQI in Shanghai in 2014, Whether has the obvious "week effect", "seasonal", "Spring Festival effect", using wavelet transform to decompose, using low frequency signal to reflect its overall change trend. Using descriptive statistics analysis and fuzzy comprehensive evaluation method to look for the primary pollutant-PM2.5. Then the correlation between PM2.5 and other pollutants is analyzed, so as to figure out three principal components among several variables. Multivariate statistical analysis of PM2.5 and the three principal components are verified through corresponding statistical tests. Finally we use the intelligence algorithm to evaluate and predict air quality. An evaluation system of air quality through the decision tree C4.5 is established and an intelligence evaluation model is also constructed. Through experiments and simulations, the prediction of C4.5 and LVQ nerve network for air quality is found to be more accurate. The cross validation of SVM algorithm are used to select the optimal penalty factor c and parameter g, and compare the effects of different kernel functions and normalized methods on prediction accuracy. With regard to the problem of SVM classifier parameter selection, based on the Particle Swarm Optimization(PSO) algorithm, we search for the regularization factor c and kernel parameter g of SVM in a global-optimization way, and make a contrast with the classification of BP nerve network. After comparing prediction accuracy of all the intelligence algorithm, C4.5 and PSO-SVM are believed to perform best in air quality classification and prediction.