Bank Bankruptcy Prediction Using Machine Learning Algorithms

At the heart of every financial system, banks play the vital role of financial intermediaries. For an economy to have sustainable growth, it must have a stable banking system. In the last decade, the financial and banking industry has witnessed a plethora of changes and developments. These have led to new challenges and greater exposure to risk. In the aftermath of the 2008 financial crisis. This scenario has brought to the fore an increased requirement to predict bankruptcy. The purpose of this study is to actualise the prediction of Spanish bank bankruptcy and compare the accuracy of four algorithms which are Logistic Regression, Decision Tree, Extreme Gradient Boosting and Feed-Forward Back Propagation Neural Network. This study also reviews the various methods of bankruptcy prediction. The findings in the study show that Decision Tree had an accuracy of 81.25% by using the Gini criterion. Logistic Regression method had an accuracy of 86.36%. Extreme gradient boosting method had an accuracy of 90.625%. Feed-Forward Back Propagation Neural Network method had an accuracy of 95.5%.