Data-driven Oil Consumption And Price Forecasting

Oil is one of the most important energy resources in the world and it has significant effects on global economic activities.High oil prices often lead to an increase in inflation and subsequently hurt economies of oil-importing countries.Low oil prices,on the other hand,may result in economic recession and political instability in oil-exporting countries since their economic development can get retarded.Besides,the oil consumption market is diversified under the influence of various factors such as oil prices and alternative energy sources,and plays an extremely important role in national energy security.Without any doubt,international oil consumption and price forecasting has become an increasingly hot issue in the fields of energy analysis and economic management.However,it still is an uphill task to enhance the prediction accuracy for oil consumption and oil prices due to the interactively driving factors and emergencies.In order to capture those factors,this paper firstly makes a comprehensive review of oil consumption and price forecasting based on the existing literature and the main factors affecting the oil market and the deficiencies of existing researches are obtained as well.Secondly,according to the influencing factors,we establish an oil consumption demand forecasting model with Google trends data and an oil price forecasting model with online news articles sentiment data.Finally,an oil price interval forecasting model with emergency-event-based expert system is constructed for major emergency events.The research is specified as follows:First of all,this paper analyzes the existing domestic and foreign literature from the perspectives of oil consumption forecasting,oil price forecasting and forecasting methods,and then summarizes the main factors affecting oil consumption and price forecasting and the deficiencies of existing research.The study shows that(1)many factors influence oil consumption and prices and lead to complex fractal patterns of the oil market.Traditional multiple regression analysis methods and statistically-based linear models cannot accurately measure and predict changes in consumption and price trends.Therefore,single variable nonlinear artificial intelligence models must be used for accurate analysis and prediction.Considering Internet public opinions' huge impacts on the oil market in the present era of big data,this article introduces online data into the oil market forecasting to verify whether Internet big data can improve the oil market forecasting precision.(2)The existing research on the impact of emergencies on the oil market mostly relies on the experience of experts and a systematic quantitative analysis method has not been established.Secondly,for the first deficiency,this paper proposed an oil consumption forecasting model based on Google trends data and a constructed oil price forecasting model with sentiment data of online news articles.Unlike those prediction models regardless of big data factors,this paper applies the typical Internet big data,i.e.online Google trends data and online news articles sentiment data,to the existent intelligent forecasting models to enhance oil consumption and price forecasting accuracy.The empirical studies show that(1)in the aspect of capturing the oil market driving factors,Google trends of"oil consumption" and online news articles sentiments of "oil price" are effective predictors for forecasting of oil consumption and oil price,respectively;(2)By introducing the useful online data,the Google trends "oil consumption" and online news articles sentiments of "oil price",the capabilities of predicting oil consumption and oil prices are significantly improved.Finally,for the second deficiency,this paper formulated an oil price interval forecasting model based on the emergency expert system.Firstly,in this system,emergency events are identified according to confidence interval methods and historical oil prices.Secondly,the corresponding emergency events are obtained by text mining technology and the influence ranges of the emergency events on oil prices are calculated.Thirdly,the oil price predictions are adjusted by the intervals affected by emergencies.The experimental results indicate that(1)the prediction accuracy for oil price intervals can be improved by introducing the emergency expert system;(2)based on the "decomposition and integration" technology,the accuracy of oil price interval forecasting is higher than that of a single forecasting model.In this paper,oil consumption,oil prices and oil price intervals are predicted based on the Google trends data,online news articles sentiment data and the expert system of emergency,respectively and the accuracy of oil consumption and price forecasting is improved,which will help policymakers in related sectors to provide corresponding decision-making reference.