Data Analytics in Improved Bankruptcy Prediction with Industrial Risk

An investigation into the causes of bankruptcy filing by borrowers and the impact it has onto the financial industries is undertaken in this work. One of the major catalysts for bankruptcy filing has been the COVID-19 pandemic that has infected the world from early 2020. With the many applications of data analytics in the financial industry, Bankruptcy Prediction Models (BPM) have seen a rise in popularity to ascertain a customer's financial health and predict if at all the customer is in potential financial distress based on selected key parameters. An in-depth review of bankruptcy prediction models has highlighted several flaws, mainly being a lack of industrial risk parameter to identify the individual borrowers that have been directly affected by their industry on a large scale as seen for the aviation industry during the pandemic. In testing, Random Forest produced the highest accuracy on real-world data. It is observed that the inclusion of the Industry Risk variable can increase the accuracy of the model, but also non-financial variables such as socio-economic status are an important contribution to the accuracy of determining if a customer is potential for bankruptcy filing.