APPLICATION OF ANT COLONY OPTIMIZATION TO CREDIT RISK ASSESSMENT

This paper presents a novel approach to solve feature subset selection problems using an Ant Colony Optimization (ACO) algorithm. ACO is one of the important naturally inspired intelligent techniques. It is based on the foraging behavior of real ants in nature. The proposed ACO is combined with a number of nearest neighbor classifiers. The resulting ACO algorithm is applied to classify credit risk using data belonging to 1,411 firms obtained from a leading Greek commercial bank. The objective is to classify subject firms into several groups representing different levels of credit risk. The results of the proposed algorithm are compared with those of others including SVM, CART, and with two other metaheuristic algorithms using tabu search and genetic algorithms, both of which use nearest neighbor classifiers in the classification phase. The results suggest that the proposed method is more accurate than others that have been tested in classifying credit risk.