Ethical transparency in business failure prediction: uncovering the black box of xgboost algorithm

Abstract

ABSTRACT This paper introduces an innovative machine learning model, built upon the robust and interpretable Extreme Gradient Boosting (XGBoost technique), to predict business failure. The study uses the VADIS financial strength indicator, from ORBIS, in a sample of 3,806 Spanish firms. The empirical analysis demonstrates that higher solvency, profitability, and reduced indebtedness correlate with a lower propensity for business failure. The model offers a powerful tool for early detection and intervention in financial distress, helping prevent insolvency. This study extends its focus to the ethical dimensions inherent in the application of machine learning algorithms. It attempts to demystify the proverbial ‘black box’ nature of such techniques, shedding light on their ethical foundations. By demonstrating the transparency and interpretability of the XGBoost model, we address the ethical concerns often associated with the ‘black box’ nature of machine learning algorithms.