An Application of Artificial Intelligent Neural Network and Discriminant Analyses on Credit Scoring

The research paper deals with credit scoring in banking system which compares most commonly statistical predictive model for credit scoring, Artificial Intelligent Neural Network (ANN) and Discriminant Analyses. It is very clear from the classification outcomes of this research that Neural Network compares well with Linear Discriminant model. It gives slight better results than Discriminant Analysis. However, it is noteworthy that a “Bad accepted” generates much high costs than a “Good rejected” and Neural Network acquires less amount of “Bad accepted” than the Discriminant predictive model. It achieves less cost of misclassification for the data set use in the research. Furthermore, if the final section of this research, an optimization algorithm (Genetic Algorithm) is proposed in order to obtain better classification accuracy through the configuration of the neural network architecture. On the contrary, it is important to note that the success of the predictive model largely depends on the predictor variables selection to be used as inputs of the model. Keywords: Scoring, Artificial Neural Network, Discriminant Analysis