Figure 6From: Gendered behavior as a disadvantage in open source software developmentRelative variable importance after randomization, normalized by original importance (\(n=100\)). Random Forest prediction with 100 datasets where 5% gender swapped AUC’s mean is 0.672 (SE:0.002), and 10% mean: 0.651 (SE:0.003). Variables indicate female-gender homophily, number of female collaborators and the number of followed females and males are the most sensitive to gender swappingBack to article page