Fig. 10 | Scientific Reports

Fig. 10

From: Optimizing imbalanced learning with genetic algorithm

Fig. 10

F1 Score comparison across synthetic data generation methods on three benchmark datasets. The proposed genetic algorithm variants (SGA, EGA, SVMGA) exhibit superior performance compared to traditional approaches, with SVMGA achieving optimal F1 scores on Phoneme (73.18%) and PIMA (69.78%) datasets, while EGA exhibits superior performance on the Credit Card Fraud dataset (84.04%).

Back to article page