Predicting short-term financial distress - An empirical comparison between Logistic Regression and Tree-based models applied to Italian companies

Abstract

Corporate Financial Distress Prediction (FDP) has been a major concern for companies in the last years. Therefore, it has been deemed necessary to implement some techniques for predicting whether or not a firm will incur into financial distress on the basis of available financial data, through mathematical, statistical, or artificial intelligence-based models. This dissertation is aimed at comparing the outcome of a specific set of machine learning models, namely tree-based methods, with the performance of a benchmark technique to predict corporate failure, namely logistic regression, because of its widespread use in the literature.