Air Pollutant Concentration Forecasting Using Long Short-Term Memory Based on Wavelet Transform and Information Gain: A Case Study of Beijing

Table 6

Comparison of prediction accuracy of different models (%).

	PM10	PM2.5	NO2	SO2	O3	CO
Hybrid model	8.84	7.59	4.83	9.69	7.80	5.96
IG-wavelet-SVR	16.39	17.23	21.29	22.36	20.75	21.55
IG-wavelet-RNN	12.60	12.28	13.51	16.96	12.21	9.48
IG-wavelet-GRU	10.09	9.92	13.50	10.10	9.78	7.75