<ul><li>This study proposes an oversampling strategy called MGS-GRF for rare event detection in binary classification on tabular data.</li><li>MGS-GRF is designed to handle mixed features (continuous and categorical variables) and exhibits coherence and association properties.</li><li>The method uses a kernel density estimator and locally estimated full-rank covariances to generate continuous features, while categorical features are drawn from the original samples through a generalized random forest.</li><li>Experimental results show that MGS-GRF outperforms other synthetic procedures in terms of predictive performances, as evaluated on both simulated and real-world datasets.</li></ul>

Harnessing Mixed Features for Imbalance Data Oversampling: Application to Bank Customers Scoring

Discover more