Evaluation of distressed enterprises using explainable machine learning: construction of an artificial intelligence algorithm to assist judges in the Belgian Enterprise Courts

01 January 2021 → 31 December 2024
Research Foundation - Flanders (FWO)
Research disciplines
  • Natural sciences
    • Machine learning and decision making
    • Natural language processing
  • Social sciences
    • Law and economics
    • Data collection and data estimation methodology, computer programs
    • Legal practice, lawyering and the legal profession
machine learning bankruptcy
Project description

The Chambers for Distressed Enterprises, operating within each Enterprise Court, have the complex task of investigating distressed enterprises. Their investigation may have important consequences for the enterprise under investigation, since it can be followed by the initiation of bankruptcy or judicial reorganization proceedings. For their investigation, the Chambers for Distressed Enterprises usually have access to the annual accounts of the enterprise under investigation. In Belgium, most enterprises have to deposit their annual accounts at the National Bank of Belgium. The enormous amount of data of all these accounts (together) may provide valuable insights on the broader phenomenon of distress and bankruptcy. However, it is beyond human capability to process all these data. The purpose of this project is to develop a machine learning tool that may assist the judges of the Chambers for Distressed Enterprises in their evaluation of annual accounts of distressed enterprises. The tool will take both numeric accounting-based and textual data of annual accounts into account, and will deliver an easy-to-interpret analysis of the distressed enterprise. This analysis will indicate how likely it is that the enterprise will encounter bankruptcy in the near future, as well as a concise summary of the main causes of distress. Ultimately, the goal is to be among the first EU countries that rely on AI assisted judicial assessments.