Improving the Quality of Financial Information Through Machine Learning

Abstract

This paper reviews previous research in order to emphasize the importance of financial information and its effects both on companies and stakeholders (owners, managers, investors, and creditors). It outlines the problems with one of the key financial reporting assumptions – the going concern assumption which is equalized to bankruptcy for the purposes of the analysis. The empirical analysis includes the creation of several machine learning models which classify companies as either “going concern” or “non-going concern” based on four financial indicators. The aim of the analysis is to provide insight on how machine learning approaches can improve financial information quality.