Digraph3 Data set

Perny's car selection problem

Data type

[0-100]-valued outranking graph

Problematics

Best unique choice problematics

Description

The problem concerns the construction of a unique best choice recommendation on the bais of the following pairwise outranking index (in %) delivered on eight decision alternatives concerning popular cars.

S a-vwgc b-r9gt c-gsax d-p305 e-tahg f-audi g-r18g h-alfa
a-vwgc - 75 70 62 0 0 0 0
b-r9gt 76 - 90 100 82 82 82 80
c-gsax 70 86 - 100 100 46 80 91
d-p305 64 65 94 - 88 22 94 74
e-tahg 33 57 93 100 - 0 80 86
f-audi 0 73 64 92 76 - 96 80
g-r18g 0 63 73 85 82 70 - 81
h-alfa 0 60 64 60 77 0 0 -

The non trivial best choice decision problem is a classic example recurrently used for presenting the Electre decision aid methods.

The RuBy best choice recommendation

The valued outranking graph admits three outranking kernels: {b-r9gt}, {a-vwgc, f-audi} and {a-vwgc, g-r18g}, and five outranked kernels: {a-vwgc, h-alfa}, {a-vwgc, e-tahg}, {c-gsax}, {d-p305} and {b-r9gt}.

To make a RuBy recommendation, we must have a look at the L-qualifications of the outranking kernels.

Choice a-vwgc b-r9gt c-gsax d-p305 e-tahg f-audi g-r18g h-alfa Irred. Ind. Dom. Abs.
{b-r9gt}
RuBy choice
24 76 24 24 24 24 24 24 100 100 76 57
{a-vwgc, f-audi} 70 30 30 30 30 70 30 30 100 100 70 0
{a-vwgc, g-r18g} 70 30 30 30 30 30 70 30 100 100 70 0
{a-vwgc, h-alfa} 74 26 26 26 26 26 26 74 100 100 0 74
{a-vwgc, e-tahg} 67 33 33 33 67 33 33 33 82 67 0 76
{c-gsax} 36 36 64 36 36 36 36 36 100 100 46 64
{d-p305} 40 40 40 60 40 40 40 40 100 100 22 60
{b-r9gt} 43 57 43 43 43 43 43 43 100 100 76 57

Despite the fact of being conjointly an outranking (76% credibility) and an outranked choice (57% credibility), decision action b-r9gt represents the most credible effective best choice. Indeed, it is the most robust recommandation in the sense that it is the only effective outranking choice with a concordance majority of more than 75%.

In case the decision maker would accept as satisfactory a global concordance of 70% and more, the RuBy best choice recommendation would become the quartet {a-vwgc, b-r9gt, f-audi, g-r18g}.

A thorough presentation and discussion of this example is given in Fodor & Roubens (1994).

Files

References


Return to main Digraph3 Data Sets