FINANCIAL DISTRESS PREDICTION USING A HYBRID MACHINE LEARNING AND NETWORK ANALYSIS FRAMEWORK

Abstract

Financial companies need to be able to predict financial crises in order to reduce risk and act quickly. Traditional models often run into problems because of how complicated it is for a lot of different financial factors to interact with unusual data. This research offers a method that uses both network analysis and machine learning to make predictions more accurate. To begin, we can find the most important financial signs using feature selection methods. After that, machine learning models such as XGBoost, Random Forest, and Neural Networks are used to sort the data into groups. Network analysis can be used for many things, such as modeling financial connections and finding trends in how problems spread. When it comes to figuring out what will happen, the hybrid method does a better job than individual machine learning models on real-world financial information. The results show that combining statistical learning with network knowledge can make predictions of financial trouble more accurate.