A Multiple Kernel Learning Approach for Air Quality Prediction

Table 11

Performance comparison for predicting the future 12 hour’s AQHI in HK.

	Accuracy	mse	wr	wf	wp
ARIMA	0.453	2.148	0.454	0.387	0.410
RF	0.559	0.806	0.559	0.554	0.551
MLP	0.597	0.66	0.597	0.598	0.603
SVC_linear	0.614	0.671	0.614	0.607	0.606
SVC_rbf	0.601	0.69	0.601	0.592	0.594
SVC_poly	0.59	0.716	0.590	0.585	0.583
SVC_sigmoid	0.267	4.992	0.267	0.113	0.071
LSTM	0.528	0.733	0.524	0.512	0.506
MKSVC	0.630	0.609	0.641	0.629	0.633