Fig. 2From: A self-inspected adaptive SMOTE algorithm (SASMOTE) for highly imbalanced data classification in healthcareA Illustrations of visible neighbors (green dots A and B) and invisible neighbors (red dot C and D). B Effects of samples generated from visible neighbors (blue dash lines) and invisible neighbors (red dash lines). Data points randomly generated between P and its invisible neighbors are likely to fall into the other class, which can mislead the classification model toward a biased decision boundaryBack to article page