Adaptive Combination Forecasting Model Based on Area Correlation Degree with Application to China’s Energy Consumption

Table 4

Forecasting results of China’s energy consumption by combination models based on different weights assignment methods in 2005~2010.

Year	Actual	ACD	APE; (%)	Entropy weight^b	APE; (%)	Reciprocal of MAPE weight method^c	APE; (%)	Optimal method^d	APE; (%)
2005	224682	221647.7	1.35	221535.5	1.4	221798.7	1.28	221584.1	1.38
2006	246270	246355.7	0.03	246367	0.04	246757.6	0.2	246270	0
2007	265583	264877.3	0.27	266339.7	0.29	264469	0.42	265361.7	0.08
2008	285000	284492.4	0.18	285979.8	0.34	284009.4	0.35	285000	0
2009	306600	308227.1	0.53	308087.3	0.49	308660	0.67	308093.8	0.49
2010	325000	328674.5	1.13	328950.7	1.22	328698.6	1.14	328743.6	1.15
MAPE (%)		0.58		0.63		0.68		0.51

bEntropy weight: weight assigned based on information entropy as adopted in [33].
^cReciprocal of MAPE method: weight assigned based on the reciprocal of each individual forecasting model’s MAPE.
^dOptimal method: the fundamental principle used to obtain the weight coefficient of combination forecasting is to make the APE minimization.