Abstract

Bike sharing systems (BSSs) are an important transportation alternative, and station distribution is a key component of these that is driven by user demand and resource constraints. Designing an effective BSS with appropriate station distribution requires a method that consists of steps structured in a flexible, parameterizable, repeatable, and organized way, based on and aligned with proven or accepted standards---particularly in resource-limited environments. This includes data-driven analysis of information relevant to BSS station design from various sources and in different formats. Models and algorithms are used to organize and examine the data, reduce redundant data, standardize factors, and find patterns that can inform the efficient design and implementation of a BSS. The algorithms and models used in the present study provide a data-driven approach to determining effective BSS station distribution in a city. Factor analysis and principal component analysis (PCA) were used as the various sources of data involved in the design of a BSS (i.e., data on traffic, demographic, and land use) can often overlap and/or have redundant data and these techniques allow minimizing superfluous data without losing relevant information. Econometric models were also used to identify the costs of pollutants, with the aim of locating stations in areas where pollution is a problem, and an emission-free BSS might be of greatest benefit. Patterns of potential users and mobility are derived from unsupervised learning algorithms. Finally, the set covering model (SCP), an optimization model for the distribution of stations, is used to define the number of stations in the city and their locations. This model's objective is to minimize costs while still satisfying user demand. Using this data-driven approach can help guide the strategic design and planning of a BSS. A case study using this method was carried out using data from the city of Cuenca, Ecuador, the third most populous city in this developing country. Cuenca is considered a mid-sized city and is a UNESCO World Heritage Site. When compared to the costly Spanish--Ecuadorian consortium that implemented the currently BSS running in Cuenca, applying the proposed data-driven approach to this real-life practical case study resulted in a 70--90\% match in the locations of stations. The consortium had to study the place of implementation in a great amount of depth and obtained a similar design to that obtained in this case study. This demonstrates the potential of the proposed method as a simple, effective, and low-cost method for the strategic planning of BSSs in small and mid-sized cities. The present study provides an affordable solution to the design of BSS station distribution for cities without many resources. Using this method, cities can take advantage of a standardized platform to define a network of stations through an established step-by-step process. The method of BSS design proposed here demonstrates three significant advantages: 1) in-depth knowledge of the area in which a BSS is to be implemented location is not required, as the design can be driven by existing data and can even be adapted to new data sources; 2) implementation is economical as this reduces the need to hire expensive expert personnel with knowledge and experience in implementing BSSs; 3) the method is versatile since it can accept input data of various kinds, which enables the adaptation of the solution to any small or mid-sized city. This method, therefore, provides small and mid-sized resource-limited cities with a simple and cost-effective method to design a BSS that can be tailored to particular contexts and can be adapted to the specific goals of BSS implementation in a given city.

Publication Date

4-26-2022

Document Type

Dissertation

Student Type

Graduate

Degree Name

Computing and Information Sciences (Ph.D.)

Department, Program, or Center

Computer Science (GCCIS)

Advisor

Pengcheng Shi

Advisor/Committee Member

Linwei Wang

Advisor/Committee Member

Rui Li

Campus

RIT – Main Campus

Share

COinS