A Multiple Kernel Learning Approach for Air Quality Prediction

Table 13

Performance comparison for predicting the future 3 hour’s PM2.5 IAQL in Beijing.

	Accuracy	mse	wr	wf	wp
ARIMA	0.471	1.208	0.471	0.472	0.474
RF	0.477	1.858	0.477	0.454	0.451
MLP	0.491	1.678	0.491	0.482	0.477
SVC_linear	0.444	2.363	0.444	0.37	0.336
SVC_rbf	0.496	1.641	0.496	0.469	0.471
SVC_poly	0.489	1.760	0.489	0.462	0.464
SVC_sigmoid	0.391	3.999	0.391	0.219	0.153
LSTM	0.391	3.999	0.391	0.219	0.153
MKSVC	0.525	0.945	0.525	0.525	0.529