Spatiotemporal Characteristics and Self-Organization of Urban Taxi Dispatch

Figure 8

The reinforcement learning model results. There are totally 3000 drivers and 1000 passengers in the systems. (a, c) ; (b, d) . The probability of taking a taxi and waiting around hotspots for (a) and (b) at the initial state is half, and for (c) and (d), it is 1.

(a)

(b)

(c)

(d)