Fig. 6

Collaborative Filtering Accuracy in Microbiome Subject Similarity Network: For each of the 121 nodes, we fit a model to the remaining 120 node network and given the node’s closest neighbors (based on network connectivity) sought to predict its 5-dimensional attribute vector. The reported error is the root mean squared error (RMSE) (denoted by \(\mathcal {E}\)) between the true attribute vector (x_i) and its predicted attribute vector \((\hat {\textbf {x}}_{i})\). The mean root squared error in the attributed SBM is 2.3, as opposed to the neighbor average, weighted neighbor average and SBM having mean errors of 2.06, 2.05, and 2.45, respectively