Modeling the Spatial Effects on Demand Estimation of Americans with Disabilities Act Paratransit Services

Abstract

A reliable method for predicting paratransit ridership is important, especially for the efficiency of the services offered. The commonly used aggregate regression model is most accurate for forecasting the total demand for regional areas such as whole counties or cities; however, it is likely to be geographically inaccurate. This paper proposes a geographical weight regression (GWR) model for predicting the demand for the types of para-transit services required by the Americans with Disabilities Act. The GWR model reflects better the characteristic of each area having its own coefficient for predictors rather than the same value throughout. The results show that trip demand increased proportionately to (a) the population size, (b) the ratio of senior citizens, (c) the ratio of people below the poverty line, and (d) the ratio of African-American riders. These results suggest that the predictive performance of the GWR model is better than that of the ordinary least squares (OLS) regression model. The GWR model is of greater value than the OLS model to researchers and practitioners, because the predictor variables are readily available from census data; this availability of data allows researchers to use the model after calibration.

The generators of paratransit trips by persons with disabilities

“…The proportion of elderly persons was found to be positive in all four models. Although some past studies (e.g., Koffman et al, 2007) did not find a positive association between the proportion of elderly persons and paratransit trips, the model results are consistent with two studies (LaMondia and Bhat, 2009;Kuo et al, 2013). The observed positive relationship between proportion of African American population and paratransit trips in the current study is consistent with Kuo et al (2013).…”

“…GLMM was used in several models instead of OLS because of the presence of spatial autocorrelation in the data. As noted in two studies in the literature review (Bearse et al, 2004;Kuo et al, 2013), because of the clustering of trip origins and destinations over space, models need to take into account spatial autocorrelation. Spatial autocorrelation has been defined as the property of random variables that are positively or negatively correlated compared to randomly associated pairs of observations when they are located at certain distances from each other (Legendre, 1993).…”

“…It may be noted that the study's findings on the relationship of persons below poverty and elderly persons with ADA paratransit trips are inconsistent with the study by Koffman et al (2007). An aspect of the study by Kuo et al (2013) that makes it more advanced than some other studies is its recognition that ordinary least squares models may be inappropriate to model paratransit trips when they are geographically clustered. Another study that focused on clustering of ADA paratransit trips was by Bearse et al (2004), which emphasized the importance of accounting for spatial autocorrelation when modeling trip demand.…”

“…In another study, Kuo et al (2013) used a geographical weight regression model by combining census data with trip data from METROlift, the ADA paratransit service for Houston, Texas, to estimate demand for ADA paratransit trips originating at home. From the modeling effort, the authors concluded that the size of population, the proportion of elderly persons, the proportion of African American persons, and the proportion of persons below the poverty level in census tract were positively associated with outgoing home-based trips.…”

Assessing the impact of different policy decisions on the resource requirements of a Demand Responsive Transport system for persons with disabilities

“…Their results demonstrate the need to model DRT operations of an entire day instead of only during a peak period, due to the existence of route duration limits in practice. To the best of our knowledge, only Deflorio (2011) and Kuo et al (2013) consider both spatial and temporal effects of demand for DRT services. Finally, Luo and Schonfeld (2012) indicate that existing models often suffer from the limitation of using manual or very simple vehicle routing algorithms (e.g.…”