Decision Support System for Sustainability Assessment of Power Generation Technologies - Efficient Decision Support Systems: Practice and Challenges from Current to Future - page 512

layers of instruments must be compiled for each problem. This method is handy when one
has to deal with problems hard to define and to assess expert opinions to be later used in
problem solving. Moreover, the method is better at rendering the processes of human
thinking than the method of logical strings. Besides being handy in finding the best solution,
the Analytic Hierarchy Process also facilitates qualitative expression of priorities with the
help of outranking tools.
The Graphical Evaluation Method is handy for visual representations of information related to
facts identified after assessment of alternatives. Graphical visualisation of information also
helps to determine the interrelations in and the structure of a phenomenon, and is useful in
comparison of alternatives with several criteria, because it helps to visualise the
interrelations between the respective criteria (Bertin, 1981; Khuri, 2002).
Sensitivity Analysis. Whereas the comparative scores and priorities of criteria are undefined
in many comparisons of alternatives, evaluations and the selected valuation techniques are
based on different premises. Since any evaluation aims to provide a decision-maker with the
best alternative or a ranking of alternatives, such uncertainties are important only in
assessment of their effect on the ranking. The decision-maker should find to what extent
(percentage) the actual values could deviate from the values in the tables for effect
evaluation or in the set of weights. The method offers the probability ranking of alternatives,
which may be used in the analysis of ranking sensitivity of alternatives considering the
overall uncertainty of the effects and priorities (Tam et al., 2006)
2. Outranking methods. Both input and output of these methods is quantitative. This group
includes multiple criteria methods of the utility theory and a number of other types: TOPSIS
( Technique for Order Preference by Similarity to Ideal Solution ), SAW ( Simple additive Weighting ),
LINMAP ( Linear Programming Techniques for Multidimensional Analysis of Preference ),
ELECTRE ( Elimination Et Choix Traduisant la Realite ), PROMETHEE I, II, MELCHIOR,
ORESTE, COPRAS, ARAS, etc. The methods in this group have a strict mathematical
foundation on axioms. They are convenient because each alternative has its utility expressed
in a quantitative form, and the comparison of values is simple. But these methods have a
drawback: quantitative measurements are prone to inaccuracies due to slips by respondents
or to other types of errors (Streimikiene & Mikalauskiene, 2009). When this group of
methods is used, the results produced by various criteria are ranked and then the rankings
are analysed. The outranking method is based on paired comparison of alternatives. All
pairs for a criterion in question must be compared. The better alternative of each pair is
determined by summing the results according to all criteria. This simple technique is used
for quantitative data. Qualitative data, if any, are interpreted as unknown quantitative
weights. The set S must be defined to include all strings of quantitative weights matching
the qualitative priority information. Sometimes one alternative will be preferred from the
entire set S , and in other cases one alternative may be preferred only from a certain part of
the set S , with preference given to other alternatives in other parts of the said set. The
distribution of weights in the set S is deemed unchangeable; the relative values of subsets in
the set S may, therefore, be interpreted as a probability that one alternative in each pair is
always preferable over the other. Probabilities are then summed to rank general alternatives
(Von Winterfeldt & Edwards, 1986).
The PROMETHEE method differs from other multiple criteria methods with its deeper
logics. The method is based on the so-called priority functions. Decision-makers may select
these functions and set their parameters themselves. The PROMETHEE method offers a
wide selection of functions to enable better reflection of the evaluator's opinions. At the