Decision Aiding
A rational approach to handling fuzzy perceptions in route choice

https://doi.org/10.1016/j.ejor.2004.04.043Get rights and content

Abstract

The purpose of this paper is to develop a heuristic way for handling fuzzy perceptions in explaining route choice behavior from behavioral point of view. A hybrid model where route choice decision making is described in a hierarchy uses concepts from fuzzy logic and the analytical hierarchy process (AHP) is proposed for making possible a more proper description of route choice behavior in transportation systems. Teodorovic and Kikuchi’s [Transportation route choice model using fuzzy inference technique, Paper presented at the First International Symposium on: Uncertainty Modeling and Analysis: Fuzzy Reasoning, Probabilistic Models, and Risk Management, University of College Park, Maryland, 1990, p. 140] fuzzy ‘if-then’ rules are adopted to represent a typical driver’s psychology for capturing essential preferences, pairwise, among alternatives that a driver may consider. The AHP is then incorporated in this model to capture the imaginary psychological process that represent underlying observable behavior to estimate drivers’ preference allotment among the alternatives. This new procedure is applied in a real world sample based on stated values of subjects. Findings show that this method provides intuitively and statistically promising results.

Introduction

Route choice plays a critical role in many transportation related problems most importantly in the congestion. A better understanding of microscopic aspects of route choice decision-making behavior will make possible to explain appropriately the phenomena behind these problems, and as a result efficient solutions to alleviate the problem. Modeling route choice behavior falls into the area of discrete choice models in traditional methods. They treat the objective values as crisp and use numerical techniques to explain the behavior based on random utility theorem. Many applications of these models are available in the literatures, among which the well-known models are the probit and logit type models (see for instance [5], [7], [18]). Random utility theory is based on micro-economic theory, which assumes that the subjective values of attributes assigned to each alternative are deterministic with random variation among them. Using the utility concept these models offer the selection of the best alternative that provide the highest utility among the choice alternatives. The random utility models explain very little about the mental process of decision makers; therefore, they are often criticized by behavioral scientists such as Gärling et al. [10]. It has been also argued that these models do not take into full consideration the vagueness resulted from decision makers’ perceptions on some attributes such as travel time. As in route choice behavior, the perceptions are usually characterized by subjectivity, ambiguity and uncertainty. In such cases, Zadeh’s [29] fuzzy logic has offered analysts a powerful tool to deal with these concepts. The lack of the ability of traditional random utility models to consider vagueness and ambiguity involved in human perception leads one to use fuzzy logic and approximate reasoning.

Despite the fact that fuzzy logic is a relatively new technique, the literature is extensive in transportation and traffic engineering. Comparatively, the applications of fuzzy logic in modeling route choice behavior somehow remained limited. Recently, Teodorovic [25] provided the state of the art of fuzzy logic systems for transportation engineering. Teodorovic and Kikuchi [26] were the first to attempt to model the route choice process using the concept from fuzzy logic. Since their procedure establishes a foundation for the suggested method, the detail of this model is explained later in this paper. Lotan [13] and Lotan and Koutsopolous [14] were also able to successfully model route choice behavior under the presence of information based on the concepts from fuzzy set theory and approximate reasoning. Akiyama and Tsuboi [2] used a multistage approximate reasoning structure to capture drivers’ decision-making process on route choice behavior. Ridwan [19] formulated a route choice model that takes into account the travelers with non-perfect-maximizing behavior considering the spatial knowledge of each individual traveler on routes. The fundamental element of his model was a concept named FiVP (fuzzy traveler preferences) which is concerned with discrete decision problems provided that pairwise comparisons between alternatives are available with inherent subjectivity and imprecision of human thinking.

Although, each theory has developed its own method, recently some models have appeared to combine these two theories to explain complex choice behaviors; such that Mizutani and Akiyama [17] proposed a practical hybrid approach based on utility and fuzzy logic model. Lee et al. [12] suggested a stepwise method for combining the randomness and vagueness uncertainty that may exist simultaneously in driver perceptions.

Proposed as the tool to capture drivers’ imaginary psychological process underlying observable behavior, the analytical hierarch process (AHP), first introduced by Saaty [20], has found a variety of decision-making applications and widely accepted as an efficient decision support method. Banai-Kashani [4] proposed the AHP as a new approach that possesses the properties of economic and psychometric approaches. He successfully developed a method using the AHP to explain mode choice behavior in urban travel demand modeling. Choirat and Seri [6] showed that the AHP is connected with psychometric choice theory inspired from results of psychological experiments. They also pointed out that from a descriptive point of view, the AHP provides a framework in which the usual decisional process of individuals can be interpreted through a decomposition in elementary units. Furthermore, the bases on which the AHP is grounded are essentially linked to the theory of preference, to economics and therefore to the theory of judgment measurement which is to psychometric.

As a pioneering work for the method proposed in this paper, Arslan and Khisty [3] developed a psychometric approach for explaining traveler behavior. They employed Weber’s psycho-physical law of 1834 for subdividing decision maker’s input space into subjectively equal subintervals. A fuzzy if-then rule base was prepared that represented human cognitive evaluations of alternatives, pairwise. The AHP was, then, utilized as a satisfying technique that represented the structure of human cognitive decision making process to estimate the preferences and therefore leading to the selection of best choice. They applied this procedure on a route choice case considering only travel time. The findings were statistically and intuitively promising. Nevertheless, the method had also some drawbacks such as uncertainties on defining a proper progression factor together with proper decision universe of discourse for each traveler to apply Weber’s rule.

In this paper, however, we proposed a new approach that avoids from those problems. We developed a rational choice behavior for explaining route choice, when the perceptions of drivers are modeled as fuzzy numbers. The AHP is suggested as the underlying drivers’ decision-making mechanism for route choice. We are not suggesting that drivers necessarily employ the AHP to come to a decision, since this could be a long process for them. However, we claim that they do this tacitly. A set of ‘if-then’ rules is prepared to represent drivers’ cognitive appraisals between alternatives, pairwise. For this, we adopted Teodorovic and Kikuchi’s [26] approach explained in their paper. Then, the AHP is incorporated into the system as a satisfying technique that can represent the cognitive decision making process to estimate the overall preference allocations among the alternatives. This new procedure is applied in a real world sample based on the stated values of subjects. The model is evaluated as how well the stated preference values can be approximated once the perceptions are modeled as fuzzy numbers.

The rest of the paper is organized as follows. Section 2 describes the structure of the suggested route choice behavior. Section 3 briefly explains the data exercised for the analysis. Then, Section 4 portrays the results in which also a comparison with the logit model is included. Finally, Section 5 concludes by summarizing some of the major findings based on discussions and points to some directions for future studies.

Section snippets

Structure of the suggested route choice behavior

Since building a hierarchy among the key attributes and computing relative worth among a set of alternatives via pairwise matrices constitute the core structure of the AHP, the basic structure of the suggested model can simply be explained in three main stages. The first stage deals with structuring a hierarchy among the factors contributing to the decision. The second stage covers the preparation of a set of ‘if-then’ rules that underlies the fundamental relationship between two fuzzy

The data

In order to justify that the method replicates human decision-making process in route choice, we exercised a real world data used by Akiyama and Tsuboi [1]. The survey was carried out at Gifu University (GU), Gifu, Japan, between origin, GU, and destination, JR, station (JRS) indicating typical commuting trips for this area. Fig. 3 schematically depicts the topography of the network.

A total of 93 subjects participated in the survey. They provided their stated perception values as triangular

Results

The hierarchical structure for this exercise is set in Fig. 4. Basically, a driver considers three factors for selecting his/her best route among the routes in his/her choice set. There needed 15 including the not fired rules to complete a pairwise matrix with respect to a single factor. Thus, for three factors 45 rules were executed for each subject to obtain the pairwise matrices.

We removed five out of 93 subjects who provided the higher preference values to the ones that offer higher travel

Conclusions

In this paper, we proposed a method that can be used to rationally handle fuzzy perceptions in route choice decision-making behavior. The perceptions were essentially defined as fuzzy numbers. We suggested a heuristic way of handling those numbers for route choice. First, route choice decision-making process was explained in a hierarchy. Then, drivers’ perceptions were categorized into regions indicating subjective relative worth of each fuzzy number to the others, pairwise. For this, we

Acknowledgement

Special thanks to Dr. Takamasa Akiyama (Gifu University, Gifu, Japan) for allowing us to use their data and to Miss Kaori Mizutani for providing the necessary information regarding the experiment.

References (29)

  • C. Choirat, R. Seri, Analytic hierarchy process, a psychometric approach, ACSEG (Approaches Connexionnistes en Economie...
  • C. Daganzo

    Multinomial Probit: The Theory and its Application to Travel Demand Forecasting

    (1979)
  • J.S. Dyer

    Remarks on the analytic hierarchy process

    Management Science

    (1990)
  • E.H. Forman et al.

    The analytic hierarchy process––an exposition

    Operation Research

    (2001)
  • Cited by (34)

    • A two-stage recreational destination choice study incorporating fuzzy logic in discrete choice modelling

      2019, Transportation Research Part F: Traffic Psychology and Behaviour
      Citation Excerpt :

      Although the most popular use of fuzzy logic is in control systems (washing machines, microwaves, air conditioning, traffic signals, engine throttle control, and so forth), it is now used in almost every field associated with reasoning. In the transportation planning field, fuzzy logic techniques have been discussed and applied to route choice modelling (Arslan & Khisty, 2006; Henn & Ottomanelli, 2006; Lee, Sugie, & Namgung, 2002; Murat & Nurcan, 2008; Quattrone & Vitetta, 2011), mode choice modelling (Holland, 2000; Kedia, Saw, & Katti, 2015; Kumar, Sarkar, & Madhu, 2013), transport project evaluation (Arslan, 2009), evacuation operation (Hsu & Peeta, 2013), container port evaluation (Yeo & Song, 2006), and so forth. Fuzzy logic works in three steps.

    • Day-to-day travel time perception modeling using an adaptive-network-based fuzzy inference system (ANFIS)

      2016, EURO Journal on Transportation and Logistics
      Citation Excerpt :

      In the last two decades, several studies have focused on the modeling of the traveler’s perception within the context of the day-to-day modeling of transportation networks. In a classification proposed by Yanmaz-Tuzel and Ozbay (2009), the learning process of travelers is categorized into three classes: (1) Reinforcement learning (e.g., see Nakayama et al. 2001; Arentze and Timmermans 2003; Avineri and Prashker 2003; Bogers et al. 2007), (2) Bayesian learning (c.f. Chen and Mahmassani 2004; Jha et al. 1998), and (3) Stochastic learning automata (e.g., in Ozbay et al. 2001). Table 1 presents a brief review on some of the perception updating models.

    • A fuzzy traffic assignment algorithm based on driver perceived travel time of network links

      2011, Scientia Iranica
      Citation Excerpt :

      Ghatee and Hashemi [15] assumed a fuzzy level of travel demand wherein a fuzzy equilibrium flow satisfies a quasi-logit formula for the network. Arslan and Khisty [16] developed a hierarchical hybrid model using concepts from fuzzy logic, and an Analytical Hierarchy Process (AHP) for driver route choice. This model estimates the driver preference allotment among different routes as alternatives.

    View all citing articles on Scopus
    View full text