Improving Financial Distress Prediction through Clustered SMOTE for Imbalanced Data

Financial distress prediction remains fundamental to identifying troubled businesses since it determines business stability along with economic forecast accuracy. The research evaluates the Synthetic Minority Over-sampling Technique (SMOTE) to correct class imbalance issues in financial distress prediction by studying its results when standardized through clustering and non-clustering approaches. The research determines how K-means clustering strengthens SMOTE by applying data balancing techniques inside separate clusters to improve model predictions for financial distress. Combining K-means clustering with SMOTE substantially improves model performance because XGBoost demonstrates the peak results, including 99% accuracy and 99% F1 score. Incorporating clustering methods helps SMOTE produce more accurate synthetic samples, achieving better predictive accuracy by improving class balance. According to these results, combining clustering methods and SMOTE demonstrates great potential for financial distress prediction in imbalanced datasets.