A multi-objective approach for the prediction of loan defaults

作者：Odeh Oluwarotimi^*; Koduru Praveen; Featherstone Allen; Das Sanjoy; Welch Stephen M

来源：Expert Systems with Applications, 2011, 38(7): 8850-8857.

DOI：10.1016/j.eswa.2011.01.096

• Fuzzy Simplex Generic Algorithm
• Neural networks
• Logistic regression
• Credit default

Credit institutions are seldom faced with problems dealing with single objectives. Often, decisions involving optimizing two or more competing goals simultaneously need to be made, and conventional optimization routines/models are incapable of handling the problems. This study applies the Fuzzy Simplex Generic Algorithm (a multi-objective optimization algorithm) in generating decision rules for predicting loan default in a typical credit institution. Empirical results show that the best indicators of default status are observed when repayment capacity and owners equity are low and the working capital is either low or high. Also, the two worst rule indicators are low repayment capacity, high owners' equity and medium working capital or medium repayment capacity, low owners' equity and high working capital. Published by Elsevier Ltd.

• 出版日期2011-7