Telecommuting choice modelling using fuzzy rule based networks

Student thesis: Doctoral Thesis


Telecommuting as an approach in transportation demand management has made the news a lot in recent years. Technology has enabled this growing trend, and more and more companies and families are taking advantages of it. Adopting telecommuting is a multidimensional decision making process that involves different aspects of life such as family, work and many more.Modelling telecommuting enables employers and employees to understand the main factors that influence on decision making about adopting telecommuting.
The role of subjective knowledge and linguistic variables cannot be ignored in human decision making process and Fuzzy Logic has proved to be a powerful tool for knowledge-based decision-making systems. Telecommuting as a multifaceted decision involves more on subjective knowledge rather that accurate numbers. Thus, fuzzy logic is applied for modelling telecommuting.
Moreover, the complex internal decision making process for adopting telecommuting reveals the role of various factors at different levels that influence on the outcome of the decision. Therefore, Fuzzy Rule Based Network, as a novel approach in modelling complex systems, is utilised. Using Fuzzy Network as a transparent approach, enables to understand the role of external inputs, intermediate variables and their interaction in modelling telecommuting.
According to choice theory and in order to find the maximum utilities of alternatives in telecommuting, the Fuzzy Network is tuned and optimised in terms of rules and membership function using Genetic Algorithm and Fuzzy c-mean clustering method. In addition, to reduce the size of Fuzzy Network, an input and branch selection method is proposed. Linguistic composition of the nodes in Fuzzy Network is also performed by an efficient method to reduce computational costs.
Results highlight the most important external and intermediate variables as well as decision rules in describing the suitability of telecommuting. Also, a Multinomial Logit model, as benchmark model, is developed to compare models performances which shows the superiority of the proposed method in transparency, efficiency and interpretability criteria.
The main contributions of this research can be highlighted in modelling the suitability of telecommuting using Fuzzy Rule Based Network, developing fuzzy utility model using Fuzzy Rule Based Network, tuning Fuzzy Rule Based Network using Genetic Algorithm, input and branch selection for Fuzzy Rule Based Network and finally proposing an efficient method for linguistic composition of Rule Based Network.
Date of AwardFeb 2017
Original languageEnglish
Awarding Institution
  • University of Portsmouth
SupervisorAlexander Gegov (Supervisor)

Cite this