Research On Methanol Price Prediction Based On Combination Prediction Model

Industry,as an important support of our national economy,determines the degree of national modernization and the power of international discourse,so it has been widely paid attention to by scholars.The chemical industry is one of the most important components of the industry,so it is also a hot academic research topic.Methanol is an important raw material for the production of formalin and the synthesis of organic compounds,as well as an important fuel.Due to various factors such as supply and demand,production costs,seasonal factors,and emergencies,Methanol prices fluctuate and have a large range of changes.Although there are some research results on predicting the price of chemical materials,they are still in the initial stage and have not formed a systematic research system.Therefore,predicting the trend of methanol prices has important theoretical and practical significance.In this paper,the price fluctuation of methanol is studied.The main research contents and conclusions are as follows:(1)The research status of empirical smoothing and combined forecast method at home and abroad is reviewed respectively.Based on the analysis and summary of the existing research results,an adaptive smoothing coefficient method is proposed,which changes the smoothing coefficient with the change in time and the number of forecast periods.It is a time prediction method,which compares the past data with the forecast data to find out the changing trend,and reflects the changing trend in the new predicted value,to achieve more accurate prediction objectives.By comparing several specific smoothing coefficients,it is found that the accuracy of adaptive exponential smoothing is better than that of traditional quadratic exponential smoothing and cubic exponential smoothing.(2)The monthly methanol price combination prediction model is constructed,and a dynamic descending sequence combination prediction model with time is proposed.The principle is to sort and assign different weights according to the accuracy of a single prediction method.In this paper,the accuracy of the established self-adaptive thrice exponential smoothing,grey GM(1,1),unitary linear regression prediction model is arranged from highest to lowest,and different weights are given in descending order.Example analysis shows that the prediction accuracy of combination prediction is as follows from high to low: ascending combination forecasting model,ascending combination forecasting model,and arithmetic average combination forecasting model.According to the forecast results,after April 2022,as the COVID-19 epidemic weakened,the price of methanol in China slowly decreased.