Traffic Status Prediction of Arterial Roads Based on the Deep Recurrent Q-Learning

Table 7

Parameter combinations for the first test.

Group	Learning rate	Reward delay	Greedy	Optimal loss coefficient
a	0.01	0.9	0.9	0.075
b	0.01	0.9	0.6	0.074
c	0.01	0.9	0.3	0.067