Resumen
Decision trees made by visualizing the decision-making process solve a problem that requires more successive decisions to be made. They are also used for classification and to solve problems usually addressed by regression analysis. One of the problems of classification that arises is the proper classification of bankrupt companies and non-bankruptcy companies, which is then used to predict the likelihood of bankruptcy. The paper uses a random forests decision tree to predict bankruptcy of companies in the Republic of Serbia. The research results show the high predictive power of the model with as much as 98% average prediction accuracy, and it is recommended for auditors, investors, financial institutions and other stakeholders to predict bankruptcy of companies in Republic of Serbia.