The hierarchical structure is a typical framework for many complex decision problems, with the criteria organized in different levels of a tree structures, whose root is the target, and whose leaves are the low-level criteria or index. This is the case of the two applications propose in this book, namely, the Vocational and the Sustainability trees. As in other decision problems, the multi-criteria (MC) analysis helps the Decision Maker, with the aim to rank the different alternatives. The most diffuse MC approaches are based on MAVT (Multi Attribute Value Theory), using a function which aggregates the criteria values into a single index (score). Among the wide number of the MAVT algorithms proposed in the specialized literature, we quote the WA approach (Weighted Averaging). Even if it is one the most popular MAVT method, it fails when interactions among criteria need to be included. For such complex problems, more general approaches are required. For this reason, we describe here the multi-linear operator, which is able to include interactions among subset of criteria. The parameters of the system are the so-called non additive measures, and are obtained by means of a questionnaire proposed to a panel of Experts. An algorithm extracts the value of the measures for each Expert, and average over all the members of the panel. Next the multi-linear method is applied to the two decision trees, showing in some cases a non linear behavior evidence.
Un metodo di aggregazione multi-criteriale per la valutazione della "vocazionalità" al riuso e della "sostenibilità" di progetti
GIOVE, Silvio
2006-01-01
Abstract
The hierarchical structure is a typical framework for many complex decision problems, with the criteria organized in different levels of a tree structures, whose root is the target, and whose leaves are the low-level criteria or index. This is the case of the two applications propose in this book, namely, the Vocational and the Sustainability trees. As in other decision problems, the multi-criteria (MC) analysis helps the Decision Maker, with the aim to rank the different alternatives. The most diffuse MC approaches are based on MAVT (Multi Attribute Value Theory), using a function which aggregates the criteria values into a single index (score). Among the wide number of the MAVT algorithms proposed in the specialized literature, we quote the WA approach (Weighted Averaging). Even if it is one the most popular MAVT method, it fails when interactions among criteria need to be included. For such complex problems, more general approaches are required. For this reason, we describe here the multi-linear operator, which is able to include interactions among subset of criteria. The parameters of the system are the so-called non additive measures, and are obtained by means of a questionnaire proposed to a panel of Experts. An algorithm extracts the value of the measures for each Expert, and average over all the members of the panel. Next the multi-linear method is applied to the two decision trees, showing in some cases a non linear behavior evidence.
I documenti in ARCA sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
