Ensemble Learning in Predicting Financial Distress of Indonesian Public Company

Predicting financial distress can avoid firm bankruptcy. That is an important issue in matters of company sustainability and the economic growth in general. Indonesia as a developing country needs a reliable system that is able to predict the bankruptcy of a company because it can affect the overall economic condition at different levels. The ensemble learning which is built to achieve better performance of prediction can be implemented to forecast the unhealthy company conditions. Random forest ensemble learning and AdaBoost have been proven superior to the single one. Both methods are applied to Indonesia Public Company data with 6 variables based on Altman Z-Score and one additional variable. The accuracy, precision, recall, and f1-score have an average of 91% regardless of the data imbalance. The ensemble score determines its superiority to the single machine learning.