Figure 2.
Depiction of data simulation and analytic process. For dense weighted networks, time series data are generated. From these, correlation and (reflected) Euclidean distance matrices are computed. Sparse count matrices are made directly using the LFR benchmark simulation program. All data generated across all conditions are then subjected to each of the 6 community detection algorithms and their performance evaluated. “V” indicates variable or node; “O” indicates the observation, and “C” the condition.