An Improved Artificial Colony Algorithm Model for Forecasting Chinese Electricity Consumption and Analyzing Effect Mechanism

Table 4

The forecasting error of these models.

Prediction methods	Year	TEC; (billion kW⋅h)	Predictions; (billion kW⋅h)	AE	APE	MAE	MAPE
The improved ABC model	2010	41923.00	41779.58	143.42	0.342%	388.95	0.928%
	2011	46928.00	46231.69	696.31	1.484%
	2012	49762.64	49455.53	307.11	0.617%
	2013	53223.00	52698.57	524.43	0.985%
	2014	55233.00	55506.47	273.47	0.495%
The quadratic regression mode	2010	41923.00	40563.41	1359.59	3.243%	1576.54	3.761%
	2011	46928.00	44276.43	2651.57	5.650%
	2012	49762.64	48187.10	1575.54	3.166%
	2013	53223.00	52295.45	927.55	1.743%
	2014	55233.00	56601.46	1368.46	2.478%
The ANN model	2010	41923.00	43212.85	1289.85	3.077%	1335.66	3.186%
	2011	46928.00	45814.30	1113.70	2.373%
	2012	49762.64	50693.92	931.28	1.871%
	2013	53223.00	54074.38	851.38	1.600%
	2014	55233.00	57725.06	2492.06	4.512%
The GA model	2010	41838.90	41870.05	31.15	0.074%	659.23	1.576%
	2011	46503.54	46059.59	443.96	0.955%
	2012	49881.91	49236.22	645.69	1.294%
	2013	53493.77	52476.52	1017.25	1.902%
	2014	56330.92	55172.84	1158.08	2.056%
The PSO model	2010	41923.00	41870.05	52.95	0.126%	450.89	1.076%
	2011	46928.00	46059.59	868.41	1.851%
	2012	49762.64	49236.22	526.42	1.058%
	2013	53223.00	52476.52	746.48	1.403%
	2014	55233.00	55172.84	60.16	0.109%