What Separates High-Performing Taxis From the Rest? A Case Study in Bangkok

Abstract

Taxi is a public transportation system widely used in many cities, including Bangkok, Thailand. It provides a convenient way for people to move around the city. However, past studies have shown that the Bangkok taxi system is highly inefficient. In fact, most taxis are vacant most of the time. Despite this fact, there are highly efficient taxis that outperform the others in terms of gaining higher revenue. Learning what these high-performing taxis do differently is essential to improving the overall system’s efficiency. In this paper, we investigate the key factors that differentiate high-performing taxis from low-performing ones. Identifying these factors allows us to draw significant insights into improving each taxi’s performance. Based on an analysis of real taxi trajectory data, this study shows that the two crucial factors that separate high-performing taxis from low-performing taxis are the passenger searching time and the ability to select less congested search routes. Thus, an effective solution needs to concentrate on improving these two factors. Enhancing other elements, such as passenger delivery time and delivery route, is much less effective and will not have any critical impact. Lastly, this paper presents an effective regression model for predicting the efficiency of an individual taxi based on its average vacant time and the proportion of time spent in congestion during a passenger search. The model performs well, with only an 8.28% error rate.