This paper focuses on modelling of Kentucky's Lexington/Fayette County specialised transportation (WHEELS) monthly ridership. The 1979 through 1985 time-series data suggest an intervention model to replicate monthly ridership. The identified model successfully incorporated the lag structure and functional forms that constitute the relationships between monthly ridership and service changes, such as service area expansion and fare increase. The selected model satisfies all estimation and diagnostic requirements. Model predictions for 1985 were quite reasonable when compared with actual ridership: cyclical patterns were correctly replicated. The supremacy of intervention modelling when compared with multiple linear regression analysis was found to be in capturing ridership seasonality, properly reflecting the impact of changes in service attributes, and displaying uncorrelated residuals.
Abstract