Fig. 1From: Towards explainable community findingExperiments for determining explainable social network analysis metrics in the node feature experiment. A similar methodology is applied to the node-pair experiment. After R runs of the algorithm, a coassociation matrix is constructed encoding how often two nodes are classified in the same community. Feature values are computed and provided as input to a random forest classifier to determine permutation importance. The distributions of permutation importance can be compared across all graphs to identify explainable metricsBack to article page