ASSESSING EFFECTS OF BUS SERVICE QUALITY ON PASSENGERS' TAXI-HIRING BEHAVIOR

Due to low quality of bus service in a congested road network, some bus-waiting travelers would take taxis instead in order to save time or get to their destinations on time. However, the correlation between bus service quality and passengers' taxi-hiring behavior is essentially unknown. This paper aims to assess the effects of bus service quality on taxi-hiring behavior based on historical data from the Global Position Systems (GPS) equipped buses and taxis in the city of Shenzhen, China. The taxi-hiring behavior is captured by analyzing the taxi-data, such as the origins of passenger pick-up, destinations of passengers drop-off, and taxi paths from the taxi movement data. The quality of bus service is assessed based on the bus location information. Parametric, semiparametric and nonparametric models are developed to explore the effects of bus service quality on taxi-hiring behavior. The results indicate that bus speed, headway and stoppage time are the core factors affecting passengers' taxi-hiring behavior. Availability of metro, time of the day and bus route directions are the secondary important factors. This study shows that when buses run with relatively low and stable speed, taxi-hiring behavior is sensitive to the slight change of bus speed. More passengers would like to take taxis when bus speed starts to decline, or speed or stoppage time of buses tends to be irregular. However, the effects of bus headway on taxi-hiring behavior are more complicated. A specific turning point (coefficient of variability of mean headway approximate to 0.7) in the relationship between taxi-hiring behavior and bus headway is shown in this paper. Based on data mining and model development, this research presents details on attributes of bus service that drive passengers to switch to taxis and how changes in those attributes encourage modal shift from buses to taxis.