The impact of payment delays on bankruptcy prediction A comparative analysis of variables selection models and neural networks

Predicting bankruptcy is important in companies' management and relates to firm's survival and development. The discriminant variables that can be selected to predict firm health influence significantly the accuracy of the models especially if we have missing data. We studied the impact of customers and suppliers payment delays to bankruptcy prediction by analyzing the relationship between neural network models and variables selection techniques (MDA, LR, DT). These variables demonstrated that they are early warning ratios to track by investors or creditors.