Table 3.

Optimum ANN architecture for the prediction of MCPA adsorption capacity.

no.	neurons	training			testing			validation
		R2	R2adj	RMSE	R2	R2adj	RMSE	R2	R2adj	RMSE
1	[3]	0.985	0.920	1.723	0.989	0.983	1.541	0.977	0.962	1.750
2	[4]	0.996	0.953	0.686	0.997	0.994	0.612	0.978	0.971	0.608
3	[5]	0.986	0.981	1.511	0.988	0.981	1.420	0.986	0.981	1.407
4	[6]	0.998	0.985	1.095	0.999	0.997	1.201	0.977	0.970	1.540
5	[7]	0.997	0.991	2.188	0.998	0.995	2.321	0.967	0.918	3.053
6	[8]	0.998	0.996	0.088	0.999	0.997	0.024	0.981	0.979	0.066
7	[9]	0.998	0.994	1.620	0.990	0.988	1.511	0.965	0.964	1.657
8	[10]	0.996	0.987	2.797	0.989	0.981	2.833	0.982	0.981	4.729
9	[5 5]	0.988	0.913	1.668	0.932	0.931	0.321	0.941	0.930	0.944
10	[5 7]	0.934	0.922	0.392	0.912	0.908	1.443	0.910	0.906	1.321
11	[7 6]	0.911	0.902	1.866	0.905	0.910	1.832	0.909	0.899	1.612