Analysis On Charateristics And Price Prediction Of Cahinese Carbon Market

Since the 21 st century,environmental problems caused by carbon emissions have become more and more serious.In order to solve this problem,many relevant policies have been formulated at home and abroad,among which the carbon trading market has the best effect.Carbon trading market controls carbon emission through market mechanism,and carbon trading price is the product of carbon trading market.In recent years,the establishment of carbon trading market is also an important policy for China to realize "double carbon" goal as soon as possible,as well as for China to bear emission reduction as an international power,and since the opening of carbon market in 2013,there have been eight mature carbon trading pilot cities.Therefore,starting from eight Chinese carbon trading pilot cities,it is particularly important to analyze the spatial similarity of Chinese carbon trading market cities and how the carbon price is influenced by other factors,as well as to establish the carbon price prediction model.Firstly,the spatial similarity of eight pilot cities of carbon trading is analyzed from a horizontal-macro perspective.In order to make up for the subjectivity of K-means clustering method,this paper analyzes urban spatial similarity in combination with systematic clustering method,and takes descriptive statistical data such as energy intensity,per capita GDP and the proportion of secondary industry,urbanization rate and carbon price as the index of urban similarity analysis.Finally,the clustering results are classified into three categories,including Shenzhen,Beijing and Shanghai,and Chongqing,Fujian,Guangdong,Hubei and Tianjin.Secondly,from the perspective of longitudinal and microscopic,the paper analyzes the influencing factors of carbon price in eight pilot cities.In this paper,Principal Component Analysis(PCA)and Pearson Correlation Coefficient(Pearson)are used to study the influencing factors of carbon price in eight pilot cities.Eight influential factors,such as oil price,natural gas price,interest rate and exchange rate,are taken as analysis indicators at energy price level,economic level,financial level and natural environment level.Firstly,Pearson was used to eliminate the factors that did not pass the significance test,and then PCA was used to eliminate the non-principal components.Finally,the absolute value of correlation coefficient and the evaluation factor of principal components were taken as the main factors affecting the carbon price of each city.Based on the analysis results of horizontal,vertical,macro and micro,this paper puts forward targeted carbon trading management guidance suggestions for three categories and eight pilot cities.Cities in the first category need to stabilize carbon markets and accelerate innovation in "green" products;cities in the second category need to accelerate industrial transformation and cultivate more qualified market participants;and cities in the third category need to improve energy efficiency and strengthen management and promotion of carbon markets.Finally,the carbon price prediction model of eight pilot cities is constructed.Taking the main influencing factors of each city obtained above as the prediction input,several common deep learning models and machine learning models with sparse self-attention mechanism are studied to realize carbon price prediction.The results show that the predictive modeling effect of machine learning with sparse self-attention mechanism in eight cities is the best,which can provide references for the selection of carbon price prediction models.