A Study Of Carbon Price Prediction Model Based On Two-stage Characteristic Selection Method And Improved ELM Algorithm

Today,carbon trading has emerged as an effective means to reduce greenhouse gas emissions due to the abnormal global climate change caused by massive greenhouse gas emissions.Since the trend of carbon trading price movement plays a fundamental role in the decision making of relevant market participants,it makes scientific carbon trading price prediction a hot issue of concern for market participants and scholars.In recent years,scholars at home and abroad have conducted a lot of researches on carbon trading price prediction models.Although the existing carbon trading price prediction models can obtain better prediction results,there are still some problems.Firstly,the great potential of machine learning models to improve carbon trading price prediction performance is not fully recognized;secondly,most prediction models do not take into account the importance of feature selection;in addition,existing studies do not consider the influence of external factors on carbon trading prices.In this paper,with the goal of accurate and efficient prediction of carbon trading price,a prediction model based on extreme learning machine(ELM)algorithm is constructed to predict carbon trading price by machine learning as well as statistical methods.Specifically,firstly,a two-stage feature selection algorithm based on Relief F and PACF is proposed to screen the eigenvalues and emphasize the importance of eigenvalues in forecasting.Secondly,the impact caused by the Russia-Ukraine conflict is added as an external factor to the input feature values of the forecasting model,as a way to improve the accuracy and objectivity of the forecasting model.Then,we propose a multi-objective optimization algorithm,the dual-group artificial fish swarm algorithm,to determine the input layer parameters of the ELM algorithm to improve the stability of the ELM algorithm.Finally,the empirical analysis proves that the prediction model proposed in this paper has certain stability and significance.The comparison analysis with different forecasting models in different dimensions proves that the feature selection method proposed in this paper,the influence of external factors and the improvement of the ELM model have significantly improved the forecasting accuracy of the model.In terms of theoretical significance,this paper proposes a carbon trading price prediction model based on machine learning,feature selection,and optimization algorithm,which can theoretically solve the shortcomings of the existing prediction models;in terms of practical significance,accurate carbon trading price prediction can help to understand the change pattern of carbon trading price and understand the fluctuation characteristics of carbon trading price,which can contribute to the establishment of a stable price pricing mechanism.In addition,accurate prediction of carbon trading price can help to grasp the dynamics of carbon trading market and provide theoretical reference for improving the cap value of carbon emissions.