Fig. 2From: Cross-validation of correlation networks using modular structureSynthetic data. The distribution of empirical correlations calculated from the planted partitions varies with the number of samples (upper left), number of modules (upper right) and planted covariance (lower left), but seemingly not with the module size (lower right). More samples, fewer modules and large within-module covariance gives a stronger signal of modular structure in the dataBack to article page