A Multiple Kernel Learning Approach for Air Quality Prediction

Table 16

Performance comparison for predicting the future 12 hour’s PM2.5 IAQL in Beijing.

	Accuracy	mse	wr	wf	wp
ARIMA	0.386	4.290	0.391	0.355	0.318
RF	0.456	1.933	0.456	0.442	0.433
MLP	0.477	1.663	0.477	0.457	0.451
SVC_linear	0.451	2.204	0.451	0.386	0.400
SVC_rbf	0.489	1.858	0.489	0.431	0.425
SVC_poly	0.478	1.851	0.478	0.452	0.449
SVC_sigmoid	0.390	4.001	0.39	0.219	0.152
LSTM	0.383	4.360	0.387	0.202	0.147
MKSVC	0.501	1.536	0.500	0.498	0.491