Short-Term Prediction of Air Pollution in Macau Using Support Vector Machines

Table 4

The tested results of SPM, SO₂, NO₂, and O₃ in different kernel models in 1-year experiment.

	Kernel
	Linear	Polynomial	RBF	Sigmoid	Wavelet
SPM
MAE	0.312	0.421	0.310	1.459	0.457
RMSE	0.440	0.628	0.433	5.526	0.625
CWIA	16.7%	40.1%	16.0%	90.0%	26.9%
RE	18.9%	24.1%	17.8%	35.6%	31.8%
SO2
MAE	0.513	0.663	0.515	13.072	0.679
RMSE	0.926	1.370	0.905	40.806	1.087
CWIA	18.4%	32.6%	17.9%	96.2%	23.8%
RE	7.4%	10.4%	8.5%	71.0%	8.5%
NO2
MAE	0.267	0.408	0.265	1.365	0.429
RMSE	0.382	0.743	0.376	5.132	0.594
CWIA	11.2%	37.8%	10.5%	98.4%	24.1%
RE	12.9%	20.5%	12.3%	36.7%	27.9%
O3
MAE	0.305	0.329	0.311	1.563	0.454
RMSE	0.403	0.434	0.408	1.921	0.579
CWIA	12.6%	15.8%	13.5%	58.3%	22.4%
RE	21.9%	25.8%	22.7%	84.4%	39.5%