REPOSITORY > RESULTS

Doctoral dissertation

New generation platform for multi-criteria qualitative decision modelling with method DEX

Author(s): Nejc Trdin (Author), Marko Bohanec (Supervisor)

Thesis defense date: 26.02.2016

Organization: MPŠ - Mednarodna podiplomska šola Jožefa Stefana

PID: 20.500.12556/ReVIS-15060

Views: 2 | Downloads: 0

Abstract

DEX is a qualitative multi-criteria decision analysis method. It supports decision makers
in making complex decisions based on multiple, possibly conflicting, attributes. The
attributes in DEX have qualitative value scales and are structured hierarchically. The
hierarchical topology allows for decomposition of the decision problem into simpler subproblems.
In DEX, alternatives are described with qualitative values, taken from the
corresponding hierarchy input attribute scales. The evaluation of alternatives is performed
in a bottom-up way, utilizing aggregation functions, which are defined for every aggregated
attribute in the form of decision rules. DEX has been used in numerous practical applications
from everyday decision problems to solving decision problems in the financial and
ecological domains.
DEX's wide use in practice revealed the need to extend it in several directions. The extensions
facilitate decision makers in more complex decision problem solving. The following
five methodological extensions were identified: support for full hierarchies, introducing numeric
attributes, using general aggregation functions, probabilistic and fuzzy aggregation
of values, and relational models.
This thesis develops and extends DEX in the identified directions. Further, each extension
is formalized in order to include it in the formalization of the extended method.
Even more, the developed formalization allows for usage of all extensions at once if in one
decision making problem. We call the extended method DEXx.
The formalized DEXx is implemented in a library, which supports basic DEX functions
together with the identified formalized extensions. The library natively supports building
DEXx models, with added extensions, and evaluating alternatives. Even more, the library
supports new utilities, which enable the decision maker to further analyse and view the
developed models.
DEXx is evaluated through four complex real-life decision making use-cases, using the
DEXx implementation. The construction and evaluation of use-cases show the added
benefits of the extensions to the decision making process. The use-cases are qualitative
assessment of public e-portals, assessment of bank reputational risk, sustainability assessment
of electric energy production technologies in Slovenia, and the model for water flows
in agriculture. The models for use-cases are developed using only native DEXx framework
of eliminating the need for additional programming or manual work.
Through the use-cases we show that identified extensions provide additional means
for the decision maker during the decision making process of the extensions increase the
expressive power of the methodology, which in turn facilitates solving of a larger class of
decision problems.

Attachments

Cite this work