A cluster-based SMOTE both-sampling (CSBBoost) ensemble algorithm for classifying imbalanced data.

Area of Science:

• Machine Learning
• Data Science
• Computer Science

Background:

• Imbalanced datasets pose significant challenges in machine learning classification tasks.
• Traditional methods often suffer from information loss or data redundancy.
• Accurate classification of minority classes is crucial in many real-world applications.

Purpose of the Study:

• To propose a novel ensemble algorithm, Cluster-based Synthetic minority oversampling technique (SMOTE) Both-sampling (CSBBoost), for improved imbalanced data classification.
• To address limitations of existing over-sampling and under-sampling techniques.
• To enhance the performance of ensemble methods on imbalanced datasets.

Main Methods:

• The proposed CSBBoost algorithm integrates over-sampling, under-sampling, and ensemble techniques (Extreme Gradient Boosting, random forest, bagging).
• It aims to mitigate redundancy from over-sampling and information loss from under-sampling.
• The algorithm employs a cluster-based approach for synthetic data generation.

Main Results:

• The CSBBoost algorithm demonstrated significantly superior performance compared to state-of-the-art competing algorithms.
• Performance was evaluated on 20 benchmark imbalanced datasets using F1 score and Area Under the Receiver Operating Characteristics curve (AUC).
• The proposed method achieved better classification accuracy and robustness on imbalanced data.

Conclusions:

• The CSBBoost algorithm offers an effective solution for classifying imbalanced data.
• It successfully balances datasets while preserving crucial information.
• The algorithm's applicability is validated through real-world dataset analysis.