Spatiotemporal DeepWalk Gated Recurrent Neural Network: A Deep Learning Framework for Traffic Learning and Forecasting

Table 3

Impact of different number of DeepWalk window size on PeMSD8 dataset.

Prediction (min)	Metrics	W = 5	W = 10	W = 15	W = 20
15	MAE	1.009	1.005	1.010	1.009
	RMSE	2.094	2.077	2.092	2.089
	MAPE	1.815	1.805	1.815	1.812
30	MAE	1.252	1.245	1.253	1.253
	RMSE	2.705	2.699	2.708	2.702
	MAPE	2.250	2.239	2.250	2.242
60	MAE	1.614	1.569	1.626	1.582
	RMSE	3.456	3.485	3.470	3.458
	MAPE	2.859	2.783	2.870	2.799

Bold shows the best.