Air-pollution prediction in smart city, deep learning approach

Journal of Big Data

Table 1 Results of different models in 1 and 7-days Lags (Bold indicates the best results.)

Model	Batch	1 Day			7 Day
Model	Batch	MAE	RMSE	\(R^{2}\)	MAE	RMSE	\(R^{2}\)
GRU	24	9.960	16.571	0.979	12.868	21.526	0.953
	32	9.541	16.609	0.978	11.055	18.122	0.974
	64	10.021	17.220	0.977	12.258	19.797	0.965
	128	9.842	16.904	0.978	11.899	19.560	0.970
LSTM	24	9.102	15.999	0.980	11.916	19.255	0.969
	32	9.503	16.217	0.980	11.623	19.154	0.972
	64	9.252	16.044	0.980	11.551	19.155	0.970
	128	9.725	16.997	0.978	11.823	19.227	0.971
Bi-LSTM	24	8.947	15.710	0.982	12.204	20.050	0.966
	32	8.868	15.597	0.982	11.253	18.488	0.974
	64	9.561	16.380	0.980	12.055	19.323	0.969
	128	9.488	16.456	0.979	11.753	18.113	0.971
Bi-GRU	24	9.907	16.859	0.978	11.488	18.854	0.969
	32	9.692	16.712	0.978	11.984	19.294	0.970
	64	9.192	16.196	0.981	11.631	19.289	0.969
	128	9.230	16.046	0.981	11.553	19.113	0.970
CNN	24	9.663	17.062	0.978	10.693	17.962	0.973
	32	9.591	16.981	0.979	11.150	18.606	0.975
	64	9.261	16.667	0.979	10.621	18.431	0.975
	128	9.974	17.636	0.977	10.674	18.636	0.974
CNN-LSTM	24	9.198	16.523	0.981	9.353	16.724	0.978
	32	6.742	12.921	0.989	9.034	16.625	0.979
	64	7.869	15.757	0.982	9.885	18.373	0.976
	128	8.940	16.337	0.980	9.037	16.524	0.979
CNN-GRU	24	9.812	17.554	0.977	9.912	18.564	0.965
	32	9.459	17.700	0.977	9.459	18.700	0.967
	64	9.433	16.836	0.979	9.653	17.856	0.976
	128	9.499	17.285	0.979	9.949	17.885	0.976