Seed germination prediction of Salvia limbata under ecological stresses in protected areas: an artificial intelligence modeling approach

Table 2 The performance measures of the best MLPs in test phase

Activation function	Training function	Structure	Accuracy	Train	Validation	Test	All
Logsig-logsig-purelin	LM	4-10-10-1	R²	0.95	0.92	0.93	0.94
			MSE	16.26	36.53	30.86	25.22
			RMSE	4.03	6.04	5.55	5.02
			MAE	2.57	3.85	3.33	3.18
Logsig- purelin	GDM	4-7-1	R²	0.92	0.90	0.90	0.91
			MSE	35.55	54.65	55.12	43.44
			RMSE	5.96	7.39	7.42	5.87
			MAE	3.65	4.11	4.14	3.95
Tansig- Tansig- purelin	BFG	4-12-12-1	R²	0.90	0.91	0.91	0.91
			MSE	54.15	43.24	44.32	42.14
			RMSE	7.36	6.58	6.66	6.49
			MAE	4.09	3.94	3.98	3.92
Logsig- purelin	CGB	4-10-1	R²	0.94	0.90	0.91	0.92
			MSE	23.34	53.33	44.54	34.45
			RMSE	4.83	7.3	6.67	5.87
			MAE	3.05	4.03	4.01	3.51
Tansig-Tansig-purelin	GDA	4-9-9-1	R²	0.92	0.94	0.93	0.93
			MSE	36.56	25.35	31.44	32.12
			RMSE	6.05	5.03	5.61	5.67
			MAE	3.91	3.05	3.43	3.47
Tansig-Tansig-purelin	CGP	4-13-13-1	R²	0.92	0.93	0.93	0.93
			MSE	35.15	30.67	30.45	31.23
			RMSE	5.93	5.54	5.52	5.59
			MAE	3.61	3.31	3.29	3.4

ISSN: 1472-6785