Application Of Data Mining Technology In Haze Forecast And Haze Economy Model In Beijing

The smog“attack”in 2013 made us aware of the haze,and also realized that it had a number of adverse effects on our lives.This has aroused widespread concern among people from all walks of life.Scholars in different fields of research are analyzing and studying the causes of haze,haze effects,and haze management from different perspectives.This has set off an upsurge in academic circles.Of course,in the research of the scholars,the purpose is to study the root causes of haze,the management of haze,and the improvement of the prediction accuracy of haze.At present,it has entered the era of big data,and data mining technology has followed great progress.With the massive data obtained from air monitoring,research and analysis of haze weather using data mining technology will yield unexpected results.At present,it has entered the era of big data,and analyzing these big data to tap the value of big data can provide scientific and accurate decision-making basis for preventing and controlling haze.Coupled with the huge meteorological data,using more effective,convenient and cheaper strategies to deal with these huge amounts of data to mine and obtain more meaningful information is now the primary problem to be solved.This paper aims to use the data mining algorithm to predict haze in Beijing and establish a haze economic model.Based on the research of many scholars,this paper analyzes and predicts the haze and haze economy in Beijing using data mining algorithms such as support vector machine,radial basis function and dynamic neural network.First of all,this paper applies the conformal interpolation method to deal with haze weather data sets,and combines the support vector machine and neural network prediction algorithm to find out the formation and evolution of haze,predicts the haze weather,selects the dynamic neural network as the main method of modeling.Secondly,based on the analysis of the current situation of haze in Beijing,this paper analyzes in detail the impact of Beijing’s haze on Beijing’s economy（agriculture,energy,industry,construction,tourism,import and export trade,etc.）And make gray correlation analysis between haze concentration and their respective variables in Beijing.Finally,the establishment of Beijing haze economic model,the use of the established model of haze governance measures.The main contributions of this dissertation are as follows:Firstly,PM_2.5 and PM10 in the meteorological data set are selected as the target factors and other attributes（SO₂,CO,NO₂,O₃）are selected as predictors.Three methods,namely RBF neural network,SVM,and time series dynamic neural network,were used to predict and analyze the haze weather and compare the prediction accuracy of the three prediction methods.At the same time,the dynamic neural network method is selected as the main method of economic modeling of haze in Beijing.Second,in view of the proposed haze economy,the gray relational analysis method was used to analyze the annual variations of the annual average value of the dependent variable PM_2.5 and the method of the selected variable,and the index of the haze economic model was selected according to the size of the correlation.By using neural network algorithm establish haze economic model.Third,based on the change of the independent variables and the forecast of the haze economic model,the suggestions of haze control on the control of the number of vehicles,the construction area,the coal pollution and the industrial pollution are put forward.Fourth,the neural network algorithm has a very powerful function,no matter the amount of data more and less,the use of dynamic neural network algorithm in time series can have a good fitting effect and prediction accuracy.