Preference modelling approaches based on cumulative functions using simulation with applications

  • Khwazbeen Saida Fatah

    Student thesis: Doctoral Thesis


    In decision making problems under uncertainty, Mean Variance Model (MVM) consistent with Expected Utility Theory (EUT) plays an important role in ranking preferences for various alternative options. Despite its wide use, this model is appropriate only when random variables representing the alternative options are normally distributed and the utility function to be maximized is quadratic; both are undesirable properties to be satisfied with actual applications. In this research, a novel methodology has been adopted in developing generalized models that can reduce the deficiency of the existing models to solve large-scale decision problems, along with applications to real-world disputes. More specifically, for eliciting preferences for pairs of alternative options, two approaches are developed: one is based on Mean Variance Model (MVM), which is consistent with Expected Utility Theory (EUT), and the second is based on Analytic Hierarchy Processes (AHP). The main innovation in the first approach is in reformulating MVM to be based on cumulative functions using simulation. Two models under this approach are introduced: the first deals with ranking preferences for pairs of lotteries/options with non-negative outcomes only while the second, which is for risk modelling, is a risk-preference model that concerns normalized lotteries representing risk factors each is obtained from a multiplication decomposition of a lottery into its mean multiplied by a risk factor. Both approximation models, which are preference-based using the determined values for expected utility, have the potential to accommodate various distribution functions with different utility functions and capable of handling decision problems especially those encountered in financial economics. The study then reformulates the second approach, AHP; a new algorithm, using simulation, introduces an approximation method that restricts the level of inherent uncertainty to a certain limit. The research further focuses on proposing an integrated preference-based AHP model introducing a novel approximation stepwise algorithm that combines the two modified approaches, namely MVM and AHP; it multiplies the determined value for expected utility, which results from implementing the modified MVM, by the one obtained from processing AHP to obtain an aggregated weight indicator. The new integrated weight scale represents an accurate and flexible tool that can be employed efficiently to solve decision making problems for possible scenarios that concern financial economics Finally, to illustrate how the integrated model can be used as a practical methodology to solve real life selection problems, this research explores the first empirical case study on Tender Selection Process (TSP) in Kurdistan Region (KR) of Iraq; it is considered as an inductive and a comprehensive investigation on TSP, which has received minimum consideration in the region, and regarded as a significant contribution to this research. The implementation of the proposed model to this case study shows that, for the evaluation of construction tenders, the integrated approach is an appropriate model, which can be easily modified to assume specific conditions of the proposed project. Using simulation, generated data allows creation of a feedback system that can be utilized for the evaluation of future projects in addition to its capability to make data handling easier and the evaluation process less complex and time consuming.
    Date of Award2009
    Original languageEnglish
    SupervisorPeng Shi (Supervisor), Jamal Ameen (Supervisor) & Ron Wiltshire (Supervisor)


    • Decision making
    • Mathematical models

    Cite this