]Analysis: 2. level: COCO-STEP
COCO-STEP Variant-Analysis
The similarity-based variant analysis module's primary goal is the support of public procurement and tender evaluation.
In that case, when there are relatively many attributes and few objects in a price/performance evaluation's OAM (so there is a big chance to find balanced states), with the special use of STEP IX, the attributes can be evaluated in multiple steps.
The essence of the analysis steps: after each of the runs, when the similarity analysis showed a balanced state (error=0), we exclude the attributes that were used and have a non-monotone mode of action (cf. noise and constant vs. real stairs). Then we repeat the same logic with the remaining attributes until the last group of attributes don't lead to classic inconsistencies. However, this may not happen ever! So, there may be totally balanced states.
We have to draw the attention to the special life situation of variant analysis: If a caller of a tender want to make sure, that the appliers thought thoroughly their proposals, then more variants can be enforced in the tender which are needed to make to compete by the appliers themselves, in frame of the tender.
Caution is advised, however, because declaring a tender to be the winner, which is supported by arguments of extreme losses of balance, is not a good idea. For example, a Trabant and a Mercedes cannot be compared by only the facts that they both have chassis, engine, and wheels, because then the good qualities of the more expensive object will be disregarded, which is indeed a distorted sense of similarity...
