Validating bankruptcy prediction by using Bayesian network model: A case from Malaysian firm

This paper provides operational guidance for validating Naïve Bayes model for bankruptcy prediction. First, researcher suggests heuristic methods that guide the selection of bankruptcy potential variables. Correlations analyses were used to eliminate variables that provide little or no additional information beyond that subsumed by the remaining variables. A Naïve Bayes model was developed using the proposed heuristic method and it performed well based on logistic regression, which is used for validation analysis. The developed Naïve Bayes model consists of three first-order variables and seven second-order variables. The results show that the model's performance is best when the method of enter is used in logistic regression which is percentage of correct is 90%. Finally, the results of this study could also be applicable to businesses and investors in decision making, besides validating bankruptcy prediction.