Research On Oil Futures Price Prediction Based On Deep Learning

With the accelerated development of economic globalization,crude oil in fossil fuels has become the focus of attention of all countries.The price of crude oil futures,the most fundamental element in petrochemical industries such as petroleum gas,gasoline,and kerosene,has a direct bearing on global political stability and economic security,making it a major focus of research in recent times.Many factors,such as international geopolitics,US dollar exchange rate,and COVID-19,can have an impact on the fluctuation of crude oil futures prices.These factors interfere with each other,interact with each other,and work together to greatly increase the uncertainty,and have a great impact on China’s economic development and people’s lives.Therefore,Accurately predicting the change of crude oil futures price is a very significant and challenging hot spot and difficult problem.The prediction of crude oil futures price is the focus of this paper.The Integrated Empirical Mode Decomposition Method,EEMD,is employed to break down time series,neural networks,text mining,and natural language processing.For example,At the conclusion,the commencement,the most expensive,the least expensive,the trading amount,the rate of increase,the selection of the oil futures price and the fusion of oil news emotion analysis to predict the oil futures price two aspects of the research.This paper’s main concentration is twofold:(1)A crude oil futures price prediction method based on EEMD-LSTM is designed and implemented.Consider that if you directly forecast the original price series,you will ignore the inherent law of the data in different periods.In this paper,the empirical mode decomposition method EEMD is used to decompose time series,and the original crude oil futures price series is processed,and the variables are obtained after decomposition.Then,each component is input into Forecasting the crude oil futures price,the LSTM model is employed.Employing the empirical mode decomposition technique EEMD,the original crude oil futures price series is processed,and the LSTM model is then employed to forecast the crude oil futures price,thus enhancing the precision of the crude oil futures price prediction and furnishing data backing for the crude oil futures investment.Meanwhile,compared with the comparison models LSTM and BP,the fitting degree of EEMD-LSTM model is the best,and the root-mean-square error and average absolute error are lower than the other two models.Therefore,Employing the empirical mode decomposition technique EEMD,the original crude oil futures price series is processed,and the LSTM model is then employed to forecast the crude oil futures price,thus enhancing the precision of the crude oil futures price prediction and furnishing data backing for the crude oil futures investment.(2)A crude oil futures price prediction model integrating oil news emotion is designed and implemented.Nowadays,numerical data such as closing price,opening price,highest price and lowest price are generally used as experimental data in the study of oil futures price change trend.With the continuous development of natural language technology,It has become possible to predict the trend of oil futures prices by analyzing oil news.Through data mining technology,the paper obtains crude oil related news in time,extracts the features of crude oil news text through natural language processing model,calculates the positive and negative emotion coefficient of news,and constructs emotion index to predict the price of crude oil futures.Finally,the crude oil futures trading data and news emotion index are input into the CNN-GRU prediction model for prediction.The experiment demonstrates that the utilization of an oil news index to forecast oil futures prices is more precise than the numerical data,such as closing,opening,highest,and lowest prices,which are used to forecast oil futures prices.