Abstract:
Three methods for preference-based ranking of nonreusable objects are described in the case when the possible results of their use are represented as pessimistic, optimistic, and most likely estimates. The methods rely on the approximation of the binary probability preference relation by binary preference relations with respect to specially designed characteristics based on the above three estimates, namely, the median, dominant, and most likely values. The methods are verified using Monte Carlo simulation. It is shown that the median and dominant preference relations ensure a relatively high degree of approximation accuracy in most cases, while the binary preference relation with respect to the most likely value leads to a considerable reduction in the accuracy of approximation.
Key words:ranking, probability, binary preference relations, nonreusable objects, interval objects specified by three points, triangular distribution, median, dominant, most likely value.
Fulltext:
PDF file (282 kB)