Search the USF Web site Site Map USF home page


Center Identification Number: 473-04

Project Title: Transit Ridership Models at the Stop Level

Principal Investigator: 
Xuehao Chu, Senior Research Associate
Phone: 813-974-9831
Suncom: 574-9831

Center for Urban Transportation Research
University of South Florida
Tampa, Florida

External Project Contact: 
Jon Ausman, Transit Planning Program Manager
Public Transit Office
Florida Department of Transportation
Phone: 850-414-4519

I. Project Objective

The Florida Department of Transportation (FDOT) has developed Transit Level Of Service (TLOS), a software package that measures transit availability and mobility. Versions 1 and 2 of the software focused on transit availability. Version 3 focused on transit mobility by measuring how much service is provided between the two stops and how long it would take to make the trip. There is a need to include in the software a predictive tool that shall determine how many people would use the service at each stop. To do this, an accurate model of ridership prediction at the stop level must be incorporated into the current version of TLOS.

The objective of this project is to develop a model for ridership prediction at individual stops. This research is bounded by the following criteria: one, it shall address fixed-route and guideway transit services provided by the Jacksonville Transit Authority within a single model of stop-level ridership; two, it shall consider boardings by direction; three, it shall consider boardings on a daily basis; and, four, it shall consider park and ride lots in ridership estimation but shall not develop a separate demand model for using park-n-ride lots. The development of this model shall be limited to input variables for which data are available either within the public domain or provided by Kittelson and Associations (K&A). This data shall include socioeconomic, travel, community, demographic, and other variables.

II. Project Abstract

CUTR shall develop a daily stop-level ridership model. CUTR shall develop a theoretical framework for the empirical model, determine data needs, formulate an empirical model, prepare data sets, calibrate the model, and validate the model. The FDOT shall select the study area. FDOT shall review and approve data needs.

This project is part of FDOT's current research efforts to include in TLOS the ability to predict daily stop-level ridership. CUTR's responsibilities through this project are experimental design (including recommending to FDOT data needs), model calibration and validation. K&A shall search and review the literature and provide both the review and the raw documents to CUTR. In addition, K&A shall collect the required data according to the experimental design and provide CUTR the data electronically in a DBF format. Finally, K&A shall incorporate the final stop-level ridership model into the current TLOS software.

CUTR shall frequently communicate with FDOT and K&A about its progress through bi-weekly teleconference calls and monthly progress reports. CUTR shall work with both FDOT and K&A to maximize the predictive power of the model. The model shall incorporate the most current understanding of stop-level ridership behavior and utilize the most detailed available stop level data.

III. Task Descriptions

Task 1 Review Literature

K&A shall search and review the literature related to transit ridership prediction. This will include the various levels of ridership prediction (including systemwide, corridor, route, route segment, and stop level). Because much of previous efforts in ridership prediction focus on the system or route level, this task may go beyond the transit literature. The issue is point-level demand in a transportation network. K&A shall provide the results of this search and review to CUTR and FDOT. In addition, K&A shall provide a copy of the documents used in this review to CUTR. K&A shall provide the review and documents within the first two months after the project begins.

Task 2 Develop Theoretical Framework

CUTR shall develop a theoretical framework, taking into account the items listed below.

1. CUTR shall address the question of stop-level boardings versus stop-level disembarkments. CUTR shall address characteristics of the stop area, characteristics of the population served by the stop, community design in the stop area, density of development in the stop area, a comparison of travel time by transit versus car from the stop area, and how this stop relates to the entire network.
2. CUTR shall address the question of direct boardings versus transfer boardings.
3. CUTR shall address the differences between home-end versus non-home-end boardings.
4. CUTR shall address the use of other modes in the stop area
5. CUTR shall address average network information between a particular stop and all other stops in the network. CUTR shall consider both fare and time of transit use.

Task 3 Determine Data Needs

CUTR, in consultation with K&A and the FDOT, shall recommend what data will be required by the theoretical framework from Task 2 in order to develop a stop-level ridership model to FDOT.

Task 4 Select Study Area

The FDOT shall select the study area. K&A shall determine data availability for the selected study area. FDOT shall approve what data will be used.

Task 5 Specify Empirical Model

CUTR shall specify an empirical model to be calibrated and validated for the study area. This specification shall be based on both the theoretical framework from Task 2 and the data availability for the study area selected in Task 4. In addition, this task shall specify the calibration and validation approaches for the empirical model.

Task 6 Prepare Issue Paper

CUTR shall document the research design reflected in the above tasks in a draft issue paper. This draft paper shall be sent to the FDOT project manager and K&A for input. In addition, CUTR may seek reviews from two to three experts outside CUTR, FDOT, and K&A. CUTR shall allow two weeks for the review process. Once receiving these inputs and reviews, CUTR shall then modify the research design and the issue paper. This finalized issue paper will serve as the blueprint for the following tasks.

Task 7 Prepare Data Sets

K&A shall provide CUTR with a single database contain all of the demographic, transit availability, community design and ridership data, sorted by stop identification number. CUTR, in consultation with K&A and FDOT, shall divide the database into two data sets for model development and validation.

CUTR, in consultation with K&A and FDOT, shall recommend data needs and the structure of the final database to be provided by K&A. FDOT shall have approval authority should CUTR and K&A not agree on data needs and structure of the database. CUTR shall document both specifications in the issue paper from Task 6. Once getting the database from K&A, CUTR shall prepare data files from it that are appropriate for the statistical software chosen.

Task 8 Calibrate Model

CUTR, in consultation with K&A and FDOT, shall use the data sets from Task 7 to calibrate the empirical model specified in Task 5, according to the pre-selected framework. CUTR, in consultation with K&A and FDOT, shall select the calibration approach that is most appropriate to the data collected. CUTR may test the calibrated model using alternative specifications of the model, including variations in the variables and in the functional form, will be tested.

Task 9 Validate Model

CUTR shall examine the prediction errors of the calibrated model using an out-of-sample validation approach. The model will be re-calibrated with one portion of the data; the other portion of the data set will then be used to examine the prediction errors. Validation may be done for alternatively estimated forms of the model. These forms can then be compared in terms of their prediction errors. The validation results may be used in the final selection of the specification of the model, including both variables and functional forms.

Task 10 Prepare Final Deliverable

CUTR shall develop draft text to be incorporated into the TLOS Users Guide that describes the formulation of the ridership model and the results of the statistical calibration and validation. CUTR shall allow two weeks for FDOT and K&A to review the draft text. Once receiving these reviews, CUTR shall then modify the text and provide K&A with the final version.

IV. Student Involvement

Students may be involved in all steps of the research process.

V. Relationship to Other Research Projects

This project is part of FDOT's current research efforts on developing and improving a set of computerized tools for transit planning in the state.

VI. Technology Transfer Activities/Peer Review

The results of this analysis will be provided to FDOT through monthly progress reports, an issue paper, and final text for the TLOT Users Guide. Any technical papers prepared for possible publication in national or international journals in transportation research shall be submitted for review and approval by the FDOT prior to publication.

VIII. Potential Benefits of the Project

The results of this research will improve a transit-planning tool, which will be used by transit agencies, local governments, MPOs and FDOT in transit service planning, regional transportation demand forecasting, congestion management, concurrency management, and community planning.


National Center for Transit Research at the Center For Urban Transportation Research University of South Florida 4202 E. Fowler Ave., CUT100 Tampa, FL 33620-5375 (813) 974-3120 (813) 974-5168 Comments: