Advanced machine learning models for real-time decision making in dynamic data environments

Usiagwu, Michael Ehiedu and Adesina, Mayowa Timothy and Chinonso, Johnson (2025) Advanced machine learning models for real-time decision making in dynamic data environments. International Journal of Science and Research Archive, 14 (2). pp. 852-865. ISSN 2582-8185

Dynamic dataenvironments presentsignificant challengesdue to their continuousevolution, highvelocity, and heterogeneity. This study explores the application of advanced ensemble machine learning (ML) models for real-time decision-making in these settings. A comprehensive methodology is employed, incorporating ensemble techniques such as XGBoost, LightGBM, CatBoost, and Random Forest to enhance decisionaccuracy, adaptability, and robustness. The research integrates real-time data processing frameworks, featuring micro- batch processing, feature engineering, noise filtering, and synthetic data balancing through SMOTE to address data imbalance and heterogeneity. Hyperparameter tuning and iterative optimization strategies, including grid search and cross-validation, are applied to improve model performance and prevent overfitting. The ensemble framework is evaluated in real-time scenarios, demonstrating its ability to process large-scale dynamic data streams with high accuracy and low latency. The findings underscore the transformative potential of these models in domains like healthcare, finance, and autonomous systems, where real-time decisions are critical.