IEEE Access (Jan 2021)
THOR 2 Method: An Efficient Instrument in Situations Where There Is Uncertainty or Lack of Data
Abstract
The present study aims to propose an axiomatic evolution of the method, called THOR 2, based on the analysis of the original algorithm. It was proposed, in the evolution, the distinction in the attribution of weights in the sum of scores as well as the multiplication of the value of the criterion weight by the fuzzy-rough index in all preference relations. This functionality allows that, in the absence of data to fill in the classification of alternatives and weights in the decision matrix, it is possible to estimate the data and assign a low pertinence value for attributing that data, thus avoiding the elimination of the alternative or criterion due to the absence of the data. In order to validate the pertinence function proposed for THOR 2, an analysis of the ranking of alternatives in three different scenarios was carried out. In this way, the scenarios were simulated in which there was an absence of data in the original decision matrix. The analysis aimed to compare the result of the ranking of the alternatives when there is no data with the situation that the decision matrix is complete (all data are available), observing the impact on the ranking of the alternatives. In all scenarios that used data estimated in conjunction with the pertinence function, the ranking was kept in line with the ranking in the initial situation. However, when it was decided to exclude the criteria, the ordering was different from the ordering in the situation of origin.
Keywords