Predicting Fine-Grained Traffic Conditions via Spatio-Temporal LSTM

Table 2

MAEs comparison of GPR, ConvLSTM, and MapLSTM.

Algorithms		Speed	Count	TT_WE	TT_EW	TT_SN	TT_NS
GPR		70.79	46.43	66.81	70.7	63.2	66.08
ConvLSTM	Train	18.78	7.18	18.75	18.27	17.77	19.01
	Test	19.27	7.32	18.71	18.11	18.14	19.29
ConvLSTM+	Train	19.44	7.13	18.75	18.59	17.82	19.46
	Test	18.91	6.89	18.71	17.96	17.85	18.93
MapLSTM	Train	18.33	5.59	16.42	16.5	16.94	18.62
	Test	18.53	7.05	16.91	17.21	17	18.57