Sağlam and Mehmet’s SMOTE with Boosting (SMOTEWB) (Sağlam and Cengiz 2022) oversampling algorithm for imbalanced data sets. SMOTEWB is noise resistant and more successful according to Matthew’s Correlation Coefficient in various data sets. It is a SMOTE-based resampling technique which creates synthetic data on the links between nearest neighbors. SMOTEWB uses boosting weights to determine where to generate new samples and automatically decides the number of neighbors for each sample. It is robust to noise and outperforms most of the alternatives according to Matthew Correlation Coefficient metric. Package includes various resampling methods to be used in imbalanced classification. Available resampling methods are:
1- ADASYN (Adaptive Synthetic Sampling)
2- BLSMOTE (Borderline SMOTE)
3- GSMOTE (Geometric SMOTE)
4- ROS (Random oversampling)
5- ROSE (Randomly Over Sampling Examples)
6- RSLSMOTE (Relocating safe-level SMOTE with minority outcast handling)
7- RUS (Random undersampling)
8- SLSMOTE (Safe-level Synthetic Minority Oversampling Technique)
9- SMOTE (Synthetic minority oversampling technique)
10- SMOTEWB (SMOTE with boosting) (Sağlam and Cengiz 2022)
More methods will be implemented.
devtools::install_github(“https://github.com/fatihsaglam/SMOTEWB”)
Sağlam, Fatih, and Mehmet Ali Cengiz. 2022. “A Novel SMOTE-Based Resampling Technique Trough Noise Detection and the Boosting Procedure.” Expert Systems with Applications 200: 117023. https://doi.org/https://doi.org/10.1016/j.eswa.2022.117023.