A Transit Ridership Model based on Geographically Weighted Regression and Incorporating Service Quality Variables
One of the long-range objectives identified in the 2020 Florida Transportation Plan for enhancing Florida’s quality of life is to increase transit ridership. For the purpose of attracting more future transit riders, the current public transportation systems are expected to expand their services and improve their reliability. In order to make wise investment decisions, it is important for transit agencies to understand what attract transit users and what improvements will yield the maximum ridership. Many factors have been identified in the past as to influence transit use. However, the effects of certain factors related to transit level of service on transit ridership have not been adequately studied, in part because of the difficulty associated with quantifying the quality measurements or lack of data.
Another problem with previous studies is that transit ridership may vary across the State of Florida and even within an urban area. A factor that determines transit use may have different degree of influence in people’s decision making from one area to another. Such spatial variations need to be accurately modeled and quantified to allow the transit agencies to make the best decisions in allocating the resource available for different areas. In a conventional regression model, the strength of correlation is determined and assumed to be the same throughout a study area. While such models will perhaps satisfy some of the needs intended for a regional planning model, it does not allow the transit properties to fine-tune their services to best meet the needs for transit in different communities.
This project involves the development of a geographically weighted regression (GWR) model to more accurately estimate the number of people taking transit to work by simultaneously considering the spatial variations existing in the Census 2000 mode to work data and by considering additional factors such as security, pedestrian environment, vehicle loading factor, etc. The goal is to quantitatively measure the effects of some of the “soft” factors and to determine to what degree different factors affect transit use in different area. This potentially will allow future FSUTMS models to implement transit model parameters that vary in different zones to more accurately reflect their influence on mode choice. Additionally, variables identified using GWR may also be used to develop quantitative measures for transit properties to evaluate the transit potentials of a community and for planners, architects, and land developers to evaluate community designs. These measures may then be implemented in a computer tool that allows developers and elected public officials to make land development decisions and supports transit development and transportation investment decisions by the local and state transportation agencies.