Research for this bankruptcy prediction has been intensified since the 1960s (Altman, 1968). Bankruptcy prediction can be formulated as a classification model. This classification models approach involving statistical methods and machine learning methods. famous reliable method is a method of Artificial Neural Network (ANN). However, in the development of classification models using this ANN there are POINTS must be handled carefully. As the parametric model, ANN's performance depends on several parameters. Perceptron ANN able to handle data features are redundant because the weights are identical will be learned during the training process. Weight to the redundant features that will be pressed so that it becomes very small. The training process for the formation of a model takes a long time. This is because the Perceptron ANN must perform a number of iterations to perform the update in order to predict the relative weight properly on all training data
Copyrights © 2018