Simulation Study on Clustering Approaches for Short-Term Electricity Forecasting

Table 6

Forecasting results for aggregate models of the entire population.

Model	MAPE (%)	r_MAPE (%)	Acc (%)	MSE	MAPE (%)	r_MAPE (%)	Acc (%)	MSE	MAPE (%)	r_MAPE (%)	Acc (%)	MSE
	Learning sample				Validation sample				Test sample
	16.05	14.77	57.60	158	16.06	15.40	54.38	75	20.17	18.82	46.43	152
	41.17	36.68	25.98	1059	52.30	44.96	29.87	917	49.19	44.30	25.60	814
ARIMA	9.48	9.01	79.71	45	19.33	18.89	45.77	103	26.28	25.08	39.00	186
	14.76	13.74	61.31	118	18.85	18.80	41.60	91	17.38	16.45	56.85	102
	14.70	13.68	61.74	118	19.96	19.91	36.85	101	17.85	16.91	56.55	103
KNN	14.64	13.39	62.12	131	14.63	14.20	60.77	56	20.41	19.41	47.02	127
RPART	15.58	14.39	59.60	133	16.55	15.48	56.61	72	20.46	19.30	49.40	153
RF	0.50	0.42	100.00	0	14.07	13.69	63.15	49	19.46	18.31	49.70	117
NNET	13.65	12.69	65.34	103	13.62	13.35	61.66	65	17.50	16.76	52.98	110
SVR	14.82	13.62	61.78	126	14.40	14.17	55.42	58	17.03	16.04	57.14	100