BANKRUPTCY PREDICTION WITH MACHINE LEARNING MODELS

Abstract

Individuals and businesses can eliminate their debts legally by applying for bankruptcy which can impact everyone involved from the investors to individual employees. Bankruptcy is also an indicator of the health of the economy. In this research, 6819 companies’ data was analyzed with 96 features for each company initially and the relationship between the features and whether the companies were bankrupt was investigated. Four machine learning algorithms were applied for the classification task: logistic regression, random forest, and XGBoost. Their performance was compared to find the best fit for the model. The accuracy of logistics regression was 0.66, of XGBoost was 0.80, and the random forest had the highest accuracy with a value of 0.93.