Characterizing paratransit travel time variability and the causes of day-to-day variation

Abstract

The study analyzes paratransit travel time variability and investigates the effects of determining factors. Data collection was done through a travel time survey onboard paratransit vehicles on a chosen route in Kumasi. Key statistical metrics were used to describe the travel time distribution (TTD) in varying departure time windows, and various distributions were fitted to describe travel time variability. The backward stepwise regression analysis approach was used to determine the predictive variables of daily variation in travel times. The TTD did not change by narrowing the departure window in the study route's outbound direction, where a typical paratransit station is operational. The Generalized Extreme Value and Burr distributions were the best fit for the dataset. Dwell time, segment length, signal delay, and the recurrent congestion index on a given segment contributed to the daily variation in paratransit travel times.