Abstract:
ELECTRE TRI is the most applicable and developed outranking based classification method in the field of MCDA. By including a large number of parameters, it provides a huge amount of information on criteria which enriches decision making process, although calculation of these large number of parameters is very time consuming and difficult task. To tackle this problem, this paper proposes a new method called NSGA-ELECTRE, by which the NSGA- algorithm learns ELECTRE TRI and elicits its parameters through an evolutionary process. The proposed method contributes to the literature by utilizing a pair of conflicting objective functions including Type I errors and Type II errors instead of using a single criterion named “classification accuracy” which used frequently in the related works. The proposed bi-objective method is applied to six known credit risk datasets. The NRGA model is used as a benchmark for validation. Computational results indicate outstanding performance of the NSGA-ELECTRE method.
Machine summary:
A rating of companies that includes financial ratios related to credit risk and the probability of bankruptcy is given in the MCDM/A literature(Yurdakul & İç, 2004; Li & Sun, 2010; Doumpos &Zopounidis, 2011; Li et al.
Also, there is comparative and statistical analysis and data-based methods in the credit / bankruptcy risk literature (Doumpos & Zopounidis, 2007; Tsolas, 2015; Mousavi & Ouenniche, 2018; García et al.
Doumpos and Figueira (2019) recently published a sorting analysis of credit ratings for companies based on ELECTRE TRI-nC, an outranking method.
e, previously classified inventory items made by DMs. We use a hybrid NSGA ΙΙ algorithm which utilizes assignment examples as training data to induce parameters of ELECTRE TRI.
Minnetti and Leone (2014) proposed a two-stage procedure for estimation of parameters of the ELECTRE TRI decision-making profiles based on a linear programming problem.
One outstanding advantage of the proposed method in classification of the credit risk is the utilization of ELECTRE TRI method as an outranking approach which considers both qualitative and quantitative criteria in taking decision makers' preferences.
In this research, we optimize parameters by combining a multi-criteria decision making technique based on outranking relations using the evolutionary NSGA-ΙΙ.
Minnetti, The Estimation of the Parameters in Multi-Criteria Classification Problem: The Case of the Electre Tri Method, in Analysis and Modeling of Complex Data in Behavioral and Social Sciences.
An evolutionary approach to construction of outranking models for multi criteria classification: The case of the ELECTRE TRI method.
A multicriteria outranking approach for modeling corporate credit ratings: An application of the Electre Tri-nC method.