Fig. 2
From: Hellinger distance-based stable sparse feature selection for high-dimensional class-imbalanced data
Skew-insensitivity of the Hellinger distance on simulated normal distributions with four scenarios corresponding to different sizes of the minority. The black line and red vertical short segments are, respectively, the mean and standard deviation of the Hellinger distance with the ratio changing from 1 to 99 based on 100 iterations