INTEGRATING MACHINE LEARNING AND NETWORK ANALYSIS FOR FINANCIAL DISTRESS PREDICTION

Abstract

Financial institutions need crisis prediction capabilities to mitigate risk and respond quickly. Traditional models sometimes run into problems due to the fact that unexpected data makes it impossible for many different financial variables to interact. This research introduces a method that integrates network analysis and machine learning to enhance the precision of predictions. To begin, we may determine which financial indicators are most important by using feature selection approaches. Next, machine learning methods such as XGBoost, Random Forest, and Neural Networks are used to categorize the data. Identifying patterns in the propagation of issues and financial connection modeling are two of the many applications of network analysis. When it comes to predicting future occurrences using real-world financial data, the hybrid method performs better than individual machine learning models. The results show that by combining network knowledge with statistical learning, financial problem estimates can be made more precise.