. 2016 Apr 6;11(4):e0152998. doi: 10.1371/journal.pone.0152998

Table 2. Optimized support vector regression (SVR) models for effluent prediction in terms of BOD, SS, NH₄⁺-N, TN and TP (all with microbial community compositions as inputs).

Water constituents	CVmse^a	c^b	g^c	Training sets		Validation sets
Water constituents	CVmse^a	c^b	g^c	mse	r²	mse	r²
BOD	0.0424	2.143	48.50	0.00652	0.935	0.00787	0.907
SS	0.0373	59.714	8.00	0.00771	0.931	0.01401	0.930
NH₄⁺-N	0.0642	2.828	45.25	0.03196	0.717	0.03851	0.412
TN	0.0424	2.462	90.51	0.00700	0.942	0.00478	0.966
TP	0.0536	4.925	128.00	0.01201	0.848	0.02475	0.824

^aCVmse: cross validation mse.

^bc: the optimized regularization cost parameter

^cg: the optimized kernel-specific parameter.

Table 2. Optimized support vector regression (SVR) models for effluent prediction in terms of BOD, SS, NH4+-N, TN and TP (all with microbial community compositions as inputs).