Design of Deep Belief Networks for Short-Term Prediction of Drought Index Using Data in the Huaihe River Basin

Table 2

The DBN results of Bengbu SPI3.

Number of input nodes	Number of first hidden nodes	Number of second hidden nodes	RMSE	MAE	Number of input nodes	Number of first hidden nodes	Number of second hidden nodes	RMSE	MAE
		5	0.6924	0.5497			5	0.6905	0.5510
		10	0.6911	0.5567			10	0.6842	0.5425
	5	15	0.7103	0.5787		5	15	0.6858	0.5473
		20	0.7006	0.5712			20	0.7138	0.5860
		25	0.8051	0.6858			25	0.7068	0.5769
		5	0.7048	0.5702			5	0.6881	0.5472
		10	0.6911	0.5506			10	0.6899	0.5479
	10	15	0.6962	0.5616		10	15	0.6954	0.5548
		20	0.7147	0.5874			20	0.7090	0.5775
		25	0.7749	0.6606			25	0.7032	0.5805
		5	0.6852	0.5453			5	0.6930	0.5545
		10	0.6926	0.5596			10	0.7165	0.5887
8	15	15	0.6969	0.5636	9	15	15	0.7300	0.6106
		20	0.7265	0.6113			20	0.7218	0.6070
		25	0.7316	0.6236			25	0.7488	0.6375
		5	0.7224	0.6001			5	0.6915	0.5516
		10	0.6923	0.5525			10	0.6962	0.5592
	20	15	0.7152	0.5940		20	15	0.7407	0.6245
		20	0.6955	0.5560			20	0.7140	0.5885
		25	0.7794	0.6679			25	0.7309	0.6136
		5	0.6935	0.5570			5	0.6919	0.5563
		10	0.6877	0.5496			10	0.7013	0.5637
	25	15	0.7037	0.5761		25	15	0.6949	0.5617
		20	0.7470	0.6322			20	0.7669	0.6566
		25	0.7174	0.5868			25	0.7214	0.6104