Abstract This study conducts a comparative analysis of different SMOTE variants, assessing their effectiveness in diverse domains. By synthesizing the findings, it provides insights into the strengths, limitations, and future directions of oversampling methods, with a specific emphasis on SMOTE-based techniques. Through an in-depth survey of research papers and articles, it explores the principles, techniques, evaluation methodologies, and challenges associated with oversampling. This review serves as a valuable resource for researchers and practitioners, aiding informed decision-making and advancements in imbalanced classification. The proposed system is composed of six integral parts: real-time data collection, data cleaning, and feature extraction, handling of imbalanced data using various methods, selection of preferred classifiers, and the utilization of a voting principle for optimal prediction. In conclusion, the system employs a multi-model classification approach to enhance the efficiency of the aquaponics ecosystem. By leveraging the power of optimal prediction based on voting, the system evaluates the performance of four classifiers using benchmark parameters such as accuracy, time, recall, and Kappa. Through this evaluation, it identifies XGBoost and Random Forest as the most effective classifiers, based on the voting principle.
Alan : Mühendislik
Dergi Türü : Uluslararası
Benzer Makaleler | Yazar | # |
---|
Makale | Yazar | # |
---|