NEURAL NETWORKS AND GENETIC ALGORITHMS FOR BANKRUPTCY PREDICTIONS

We are focusing on three alternative techniques - linear discriminant analysis, legit analysis and genetic algorithms - that can be used to empirically select predictors for neural networks in failure predictio n. The selected techniques all have different assumptions about the re lationships between the independent variables, Linear discriminant ana lysis is based on linear combination of independent variables, legit a nalysis uses the logistical cumulative function and genetic algorithms is a global search procedure based on the mechanics of natural select ion and natural genetics. In an empirical test all three selection met hods chose different bankruptcy prediction variables. The best predict ion results were achieved when using genetic algorithms. Copyright (C) 1996 Elsevier Science Ltd