Research Article

Rebalancing Docked Bicycle Sharing System with Approximate Dynamic Programming and Reinforcement Learning

Table 3

Key performance indicators by strategies.

StrategyAverage unmet demand (person)Time (s)

Do nothing: no rebalance19.00ā€”
Strategy 1: all stations13.672,825.9
Strategy 2: near stations19.672,113.0
Strategy 3: stations with large errors14.002,021.6