Forecasting energy demand of China using Bayesian Combination model

Abstract

To analyze the impact of the related economic factors on China's energy demand, Path analysis is used to analyze the major factors and their direct and indirect effects on energy demand. This study showed that the main factors that affect the energy demand are the economic growth, the total population, and the primary energy structure, the economic growth is the main determining factor, and the primary energy structure is the major restrictive factor. On this basis and considering the multicollinearity and the validity of the forecast, we established a partial least-square (PLS) and the trend extrapolation prediction model, and then we sum up all the information to found a PLS—trend extrapolation combination forecasting model based on the optimized combining forecast theory. Finally, we obtain the probability distribution of the error using the Bayesian statistic theory and find the confidence interval of combining forecasting result. The results indicate that the outcome of combining forecasting will be more precise after using the Bayesian error correction approach.