Inferring Passenger Denial Behavior of Taxi Drivers from Large-Scale Taxi Traces

Abstract

How to understand individual human actions is a fundamental question to modern science, which drives and incurs many social, technological, racial, religious and economic phenomena. Human dynamics tries to reveal the temporal pattern and internal mechanism of human actions in letter or electronic communications, from the perspective of continuous interactions among friends or acquaintances. For interactions between stranger to stranger, taxi industry provide fruitful phenomina and evidence to investigate the action decisions. In fact, one striking disturbing events commonly reported in taxi industry is passenger refusing or denial, whose reasons vary, including skin color, blind passenger, being a foreigner or too close destination, religion reasons and anti specific nationality, so that complaints about taxi passenger refusing have to be concerned and processed carefully by local governments. But more universal factors for this phenomena are of great significance, which might be fulfilled by big data research to obtain novel insights in this question. In this paper, we demonstrate the big data analytics application in revealing novel insights from massive taxi trace data, which, for the first time, validates the passengers denial in taxi industry and estimates the denial ratio in Beijing city. We first quantify the income differentiation facts among taxi drivers. Then we find out that choosing the drop-off places also contributes to the high income for taxi drivers, compared to the previous explanation of mobility intelligence. Moreover, we propose the pick-up, drop-off and grid diversity concepts and related diversity analysis suggest that, high income taxi drivers will deny passengers in some situations, so as to choose the passengers’ destination they prefer. Finally we design an estimation method for denial ratio and infer that high income taxi drivers will deny passengers with 8.52% likelihood in Beijing. Our work exhibits the power of big data analysis in revealing some dark side investigation.

The Influencing factors of Female Passenger Background in Online Transportation with Perceived Ease of Use

“…Meanwhile, male users were only 31%. The similar result was also reported by (Surjandy et al, 2018b;Zhang & Wang, 2016). That female users were higher than male.…”

“…However, OTA also has a negative impact or risk for passenger and driver (Gao & Chen, 2019;Sarriera et al, 2017;Surjandy et al, 2018a). For example, the driver refuses passengers' orders due to skin color or being diffable (Zhang & Wang, 2016).…”

Demand, Supply, and Performance of Street-Hail Taxi

“…We assume that taxi drivers do not deny hailers, so if a vacant taxi enters the segment while at least one hailer is waiting, the driver will pick up a hailer. Note that although passenger denial exists in reality, the proportion of denied passengers is not significant [42]. This assumption is only needed to preserve homogeneity among drivers, so its impact is further weakened if taxi drivers have similar denial patterns.…”

Drivers learn city-scale dynamic equilibrium

“…Other driver decisions can affect income as well, such as driving safety [14], when to start and stop working [15], [16], detours and overcharges [17], [18], and passenger denial [19]. But based on the results of [13], we can focus on delivery speed, search strategy and experience.…”