Financial Distress Prediction Using a Network-Informed Machine Learning Framework

Abstract

The Financial Distress Prediction web application is a Django-based platform designed to
assess the likelihood of financial instability in companies using machine learning. This system
enables both users and administrators to interact through a secure registration and login
interface, offering real-time predictions based on eleven critical financial ratios. Utilizing a
dataset of 5,002 companies, the application employs a Random Forest Classifier with 100
estimators to perform binary classification, identifying whether a company is financially
distressed (1) or stable (0). The model is trained using an 80/20 train-test split and evaluates
its performance with key metrics such as accuracy, confusion matrix, and classification
report. The trained model is serialized using joblib and integrated into the system for fast and
accurate online predictions. This predictive system supports decision-making by enabling
users to input financial indicators and receive immediate insights on potential financial
distress. Administrators have additional capabilities to manage user access and monitor
application usage. The system features a user-friendly web interface for data entry and result
display, as well as data visualization tools for reviewing the dataset’s structure. By leveraging
the power of machine learning and intuitive web design, this project bridges the gap between
financial analytics and accessible decision-support systems, aiming to assist investors,
auditors, and financial professionals in proactively identifying at-risk companies.