An Empirical Study of the Support Vector Machine Model of Demand Forecast for Transportation Energy

The transportation energy demand system is a complex, multivariable, and nonlinear system. The time series data of China's transportation energy is of high instability and non-linearity. Therefore, the model of demand forecast for transportation energy established with the linear method faces a serious challenge. In this case, the non-linear model has attracted people's widespread attention. The new achievement of Machine Learning Theory — Support Vector Machines (SVM) Theory, as a non-linear modeling method, has the good non-linear quality and the high fitting precision. It has been widely applied to the modeling and the analysis for the time serious data in the economic field. In this paper, we set up a model of the demand forecast for China's transportation energy with the SVM Theory from the angel of non-linear characteristic of the time series data of transportation energy and carries on an empirical research. The model based on the SVM Theory is compared with the model on the basis of Error Correction. The result shows that the former model has higher forecast precision than the latter.