A small-sample hybrid model for forecasting energy-related CO2 emissions

Mitigating the impact of developing countries to global climate change has become an important issue in the fields of science and politics. This study proposes a small-sample hybrid model for forecasting the energy-related CO2 emissions of developing countries. The CO2 emissions of these countries have not reached the inflection point of the long-term S-shaped curve and usually present short-term linear or approximately exponential trends. This concern is considered in the design of a hybrid forecasting equation combined by a nonhomogeneous exponential equation and a linear equation. The estimated parameters of the hybrid equation are obtained by minimizing the residual sum of squares and solving for a non-constrained optimization equation. To evaluate the performance of the hybrid model, the traditional linear model, GM (grey model) (1, 1), and the hybrid model are used to forecast the CO2 emissions of China from 1992 to 2011. Analysis of forecasting results shows that the hybrid model can respond more quickly to changes in emission trends than can the two models because of the specialized equation structure. Overall error analysis indicators also show that hybrid model often obtains more precise forecasting results than do the other two models.