Short-Term Wind Speed Prediction Using EEMD-LSSVM Model

Table 7

The percentage improvements of EEMD-LSSVM over other models (Case Two).

Prediction horizon	Indicators (improvement)	Compared models
		LSSVM	ARIMA	BP	EMD-LSSVM	EEMD-ARIMA
One-step	RMSE	63.77%	64.70%	65.17%	35.25%	0.89%
	MAPE	68.20%	67.69%	69.82%	28.45%	13.47%
	MAE	65.73%	66.02%	67.04%	35.45%	2.00%
	CC	1.86%	1.99%	2.05%	0.38%	0.00%
	SSE	86.87%	87.53%	87.86%	58.06%	1.74%
	SDE	63.80%	64.63%	65.21%	35.30%	1.30%
Two-step	RMSE	67.34%	68.63%	68.32%	36.64%	27.73%
	MAPE	68.41%	68.77%	68.56%	25.50%	16.16%
	MAE	66.00%	67.14%	66.64%	34.41%	19.66%
	CC	4.78%	5.21%	5.14%	0.79%	0.36%
	SSE	89.34%	90.16%	89.96%	59.85%	47.77%
	SDE	67.47%	68.52%	68.38%	35.84%	27.46%
Three-step	RMSE	65.36%	66.53%	66.06%	33.19%	18.03%
	MAPE	69.25%	69.56%	68.59%	28.21%	26.79%
	MAE	64.72%	66.09%	65.14%	33.80%	18.79%
	CC	7.87%	8.36%	8.18%	1.14%	0.47%
	SSE	88.00%	88.80%	88.48%	55.36%	32.80%
	SDE	65.66%	66.25%	66.11%	31.69%	18.32%