Survey Efficiency Development with Fuzzy AHP Method



Survey Efficiency Development with Fuzzy AHP Method

Authors

Domokos Esztergár-Kiss, Budapest University of Technology and Economics (BME)

Description

The applicability of Fuzzy AHP method for weight calculation of multimodal journey planner analysis was investigated. Based on results of an online survey, weights were determined and compared using the AHP and the Fuzzy AHP method.

Abstract

The analytic hierarchy process (AHP) was developed by Saaty and is one of the most widely used methods, as it can solve complex decision problems. The AHP method is based on a set of simple pairwise comparisons, and from that the final weights for the aspects can be calculated. However this method is often criticized for its uncertainty and imprecision associated with the mapping of users’ perception. The fuzzy AHP approach can tackle qualitative problems of the pairwise comparisons, which provides a more efficient decision making process which provides in terms of uncertainty of human behaviour. Mikhailov presented a fuzzy method, which can derive priorities using fuzzy pairwise judgments, and also implemented a programming method for this. The fuzzy AHP method was not really used for evaluation of multimodal journey planners from the user perspective.

In order to evaluate multimodal journey planners different such aspects were defined, which are potentially important for passengers. They were classified into 5 main aspects, as route-planning services, booking and payment, handled data and operational features, comfort service information, supplementary information.
The users were assigned to 5 user groups: student, worker, tourist, businessman and pensioner. The definition of the user groups was based on their age (younger, middle aged, older), their motivation of their travel (school based, work based, leisure based), and their possible difficulty in travel (handicapped, without problem).
A survey referring to the aspects was created. The first part contained questions about the users’ age, their occupation (student, worker, pensioner), their travel situation (whether they are handicapped or not), and the reason why they use journey planners (work, leisure, tourism). According to these data it was possible to identify to what user group the participants belonged. In the second part the users were asked to rank the main aspects (route-planning, booking and payment, handled data, comfort service information, supplementary information) according to how important these services are in general for them. The rest of the survey contained questions of aspects of the journey planners. The users had to provide a value between 0 and 10 according to the subjective importance of the single aspects. The research was conducted online, 133 participants filled in the survey.

The AHP is a very flexible and efficient method, because the final results are obtained on the basis of the pairwise relative evaluations of both the aspects and the options provided by the user. The Fuzzy AHP method is a more advanced analytical method, which is developed from the AHP. The fuzzy comparison ratios should be able to tolerate imprecision and overcome uncertainties of AHP. In this case the AHP and the Fuzzy AHP methods are used only to calculate the weights of the main aspects regarding the user groups.
The AHP method generates a value for each main aspect according to the pairwise comparison results. This calculation process is performed for each user group separately. In order to compute weights for the main aspects, following the AHP method, a value matrix A is created, which contains the AHP values. The survey did not contain questions for the pairwise comparisons, therefore the AHP value was calculated pairwise from the proportions of the examined main aspects. Once the matrix A is built, it is possible to derive from it the normalized pairwise comparison matrix. Finally, the weights for the main aspects are calculated.

As Fuzzy AHP provides smoother results the calculation of weights were also processed with this method. The same AHP values were used to provide a good basis of comparison of the AHP weights and Fuzzy weights. The first step of converting AHP values is the fuzzification of the original AHP elements, which generates the fuzzy elements. From the fuzzy elements fuzzy AHP value matrix can be generated. The calculation of the weights is similar to the AHP method, however performing fuzzy operations is more complex, than in the case of AHP.

Using the AHP and Fuzzy AHP method helps to define weights of main aspects for each user group. Finally the AHP weights and Fuzzy weights were calculated and compared to the original weights. The calculations were performed by using MatLab software. In general it can be observed that the AHP generated more divergent weights compared to the other two possibilities, while Fuzzy weights resulted in a smoother distribution among the main aspects. The Fuzzy weights represent closer results to the original weights, therefore this method is better applicable for our purposes.

Publisher

Association for European Transport