The Problem of Bikeshare: Improving Rebalancing Operations for Citi Bike in New York City

Abstract

The ability to transport people efficiently is an important aspect of any city, but traffic congestion and inefficient transportation are still a large issue in major urban areas throughout the world. One solution to this problem that is gaining popularity is bikeshare. Bikeshare is convenient and self-serving, and its integration in already existing public transportation systems enhances its feasibility. New York City in- troduced its bikeshare program Citi Bike in May 2013, and it has been successful. However, as demand for service increases, new problems have arisen. Currently, their biggest issue is that of rebalancing, where there is an imbalance in the distribution of bikes leading to either full or empty stations due to fluctuations in demand at different times of the day. First, analysis of historical trip data is used to generate and predict patterns of demand and supply of bikes at each station location. These projections are then used to implement a k-means method of rebalancing in determining how to optimally reallocate the bikes within the system when necessary. Finally, this study explores possible ways for the system to self-balance through its users without having to recruit outside help to manually move bikes from one station to another. In order to achieve this process, this thesis adopts a method motivated by theories of financial market and hedge funds, and also suggests a thoughtful method in increasing Citi Bike's geographical area of operation.