Analysis of effect of information in application-based taxi service system

Abstract

With the advance of smartphone technology, application-based taxi service has occurred and many passengers and taxis are using this service. The operation of the current service follows the following procedure: the origin and destination information of a passenger are sent out to taxis

once a taxi accepts a service request, the dispatch result is informed to passenger. While there are various advantages, the destination information of the passengers that have not been disclosed before is exposed to the taxi, and there is a concern about the refusal to ride which occurs when the taxi selectively accepts the passenger. Therefore, in this paper, we study the effect of passenger’s destination information on the taxi service system, especially the change and correspondence of taxi fleet, government and passenger.

First, we want to know if the passenger's destination information in the present situation is likely to cause the refusal of boarding. We model taxi fleet that try to maximize profits, reflecting the general tendency of taxi fleet to choose passengers. In order to represent the characteristics of taxi service, we model it as closed queue network. Since it is difficult to obtain an exact solution, we derive an approximate solution by linear programming which is made by approximate mean value analysis and piece-wise linear approximation. We identify the impact of destination information by comparing the profit of taxi fleet when accepting passengers as much as possible to maximize profit and when accepting all passengers. To understand the performance of the results, we develop a reference model and compared the results. We analyze the current situation and conduct sensitivity analysis through taxi service data of Seoul City.

Second, the government should try to alleviate the problem if it is judged that there is a possibility of frequent refusal to ride in the present situation. Because the taxi fleet refuses to ride as a result of their efforts to use information to increase profits, the government adjusts rates to ensure that accepting all passengers will yield the best results. Therefore, in this study, applying the inverse optimization problem to the taxi fleet model, we derive the fare levels that will result in optimal results accepting all passengers. Differential fare structure is derived for each departure and destination of the passenger, and it is suggested that the taxi fleet will exchange the commission or incentive corresponding to the passenger information directly with the taxi fleet whenever the taxi fleet handles the passenger.

Third, the incentive structure found through inverse optimization is an incentive structure for creating a specific situation among many situations in which refusal of boarding does not occur. In this study, we try to find the incentive structure that creates the situation where the incentive amount of government per unit time is the smallest among many situations in which boarding rejection does not occur. We establish a bi-level problem between government, an upper-level decision maker who wants to reduce boarding rejection and minimize incentive amount, and taxi fleet, a lower-level decision maker that selectively accept passengers to maximize profits. Mixed integer programming and heuristic algorithms can be used to find better incentive structure. If too much incentive amount is needed to completely eliminate the riding rejection, we suggest a plan to find an incentive structure that alleviates the condition so that the minimum passenger probability is above a certain level.

Finally, we test the results of existing mathematical models with many assumptions in a more realistic environment. We develop a simulation model that relaxes assumptions about passengers and confirm changes in the value of destination information. We also identify changes in the percentage of taxi passengers by setting a discrete choice model to express passengers affected by riding rejection and price changes.

In this paper, we have examined the problems in various perspectives to understand the social impact of the disclosure of passenger destination information due to the introduction of application-based taxi service. We contribute to the identification and preparation of possible problems.