DEMAND FORECASTING AND THE AMERICANS WITH DISABILITIES ACT: ORANGE COUNTY, CALIFORNIA, TRANSPORTATION AUTHORITY'S ACCESS PROGRAM

Document Type

Journal Article

Publication Date

2004

Subject Area

ridership - forecasting, ridership - forecasting, ridership - demand, mode - paratransit

Keywords

Variables, Travel models (Travel demand), Travel demand, Time series analysis, Scenarios, Ridership, Projections, Patronage (Transit ridership), Paratransit services, Orange County Transportation Authority, Orange County (California), Forecasting, Dial a ride, Archived data, Americans with Disabilities Act, Accuracy

Abstract

A sound empirical basis for forecasting public transit's complementary paratransit ridership is critical as these programs grow and develop budgets that threaten to eclipse fixed-route operations, even as court interpretations of the Americans with Disabilities Act (ADA) narrow the latitude of operators. A reliable demand model that can be adapted to reflect change is a powerful planning tool in the evolution of ADA complementary paratransit services. Orange County Transportation Authority (OCTA) in California sought a model to provide a forecast of Access service ridership over 5 years. The model was developed from extensive time series analyses of archived records of 1.8 million daily boardings over a 34-month period within a larger study effort that included a peer survey, selected census data analysis, and community outreach. This model predicts OCTA's daily ADA one-way passenger boardings. Daily predictions can be aggregated to generate forecasts monthly, quarterly, or yearly. The current model with 15 predictor variables provides an extremely close fit to actual ridership, offering a balance of accuracy and practicality. This model has value to practitioners and researchers alike because the key variable of average daily ridership is readily available and enables the model's use and validation in a range of public transit settings. Also, the model can predict variables such as revenue hours and miles for given peak periods. Three methods are presented for adapting the model to accommodate either temporary or permanent change. The model's basic input is daily ADA ridership numbers for a 2- or 3-year period. The model's output represents not latent demand, but demand likely to present for services.

Share

COinS