Augmentation in Financial Data Evaluation Using Dimensionality Reduction and Adaptive Ensemble Learning

Abstract

The classification of financial datasets presents unique challenges due to their high dimensionality, imbalanced nature, and complexity. This research proposes a novel three-phase methodology to address these challenges effectively. In Phase-I, baseline evaluations of five machine learning classifiers—Random Forest, Gradient Boosting, Support Vector Machine (SVM), XGBoost, and LightGBM—were conducted on the original dataset. Phase-II introduced High Dimensionality Reduction with Forward Feature Elimination (HDFE), reducing irrelevant features and improving model performance. In Phase-III, a Hybrid Reverse Binary Optimization with Adaptive Fusion (HRBOAF) framework was implemented, achieving a 25.35% reduction in features and enhancing model interpretability. After hyperparameter tuning, ensemble methods (XGBoost and LightGBM) emerged as top-performing algorithms, achieving 94.0% accuracy with significant gains in sensitivity and F1-score. The findings underscore the importance of dimensionality reduction, feature selection, and hyperparameter optimization in financial data classification, offering a scalable and efficient solution for predictive modelling in complex datasets.