Page 209 - Materials Chemistry, Second Edition

10.2 MCDM methods overview 207
qualitative and quantitative data. Contrary to input-oriented, the process-oriented area could
be divided into compensatory and noncompensatory approaches (Hwang and Yoon, 1981).
Compensatory approaches allow trade-off within available attributes. In these methods, un-
favorable scores or disadvantage of an attribute can be counterbalanced by a favorable score
or advantage of another attribute. The compensatory approach can be further segmented into
three subcategories, such as:

a. Scoring approaches: In this type of approach, all available attributes are considered at
once, and an alternative with maximum utility or score is selected. An example of this
approach is simple additive weighting (SAW).
b. Compromising approaches: The approach involves selection of alternative which has the
minimum distance from the ideal solution and maximum distance from nonideal
solution; for example, technique for order of preference by similarity to ideal solution
(TOPSIS).
c. Concordance approaches: Selection of alternative in this type of approaches is based
on arranging a set of ranking preferences, which satisfies an adopted concordance
measure. An example is elimination et choixtraduisant la realit e or elimination and choice
expressing reality (ELECTRE).
Contrary to compensatory approaches, noncompensatory approaches do not allow trade-off
within the attributes and comparison of alternatives is based on considering each attribute in-
dividually. A few methods that are based on noncompensatory approaches are lexicographic,
elimination by aspects, maximax, maximin, disjunctive constraint, conjunctive constraint, and
dominance. The compensatory approaches are cognitively more challenging for decision-
makers than noncompensatory approaches; however, the results could be more optimal
(Yoon and Hwang, 1995). Segregating MADM methods into two major approach groups
suggested by Hwang and Yoon (1981) is one of the ways of classification, whereas, there are
multiple ways in which different studies have attempted to classify MADM methods. Readers
can refer to Chen and Hwang (1992), where taxonomy of the MADM methods is provided.
Triantaphyllou (2000) suggested that MADM methods can also be classified corresponding
to decision-makers, such as methods involving single decision-maker and group decision-
makers. A detailed understanding of group decision for single decision can be found in Kalbar
et al. (2013). Similarly, Kahraman et al. (2015) classified MADM methods into outranking,
distance-based, and pairwise-comparison based. Outranking approaches provide outrank re-
lationships but not any value function, whereas, distance-based methods are a development of
distance matrixes and pairwise-comparison methods compare a pair of alternatives or indica-
tors at a time in sequence. Liou and Tzeng (2012) presented an overview of MADM method
development from 1738 to 2012. The study divided MADM methods into three major catego-
ries, which are based on approaches of evaluation, weighting, and normalization.
One of the methods, named TOPSIS, has shown better performance in many applications.
TOPSIS has been shown to take into account weights more effectively (Rafiaani et al., 2019;
Kalbar et al., 2017a). Another advantage of TOPSIS is that the method takes into account the
nature of the indicators (i.e., whether the indicators are “benefit” type or “cost” type) while
processing the indicators score by creating sets of a positive ideal solution (PIS) and a negative
ideal solution (NIS). Such an approach resembles human thinking and makes it unique
among other available methods (Yadav et al., 2019; Kalbar et al., 2012). Considering the