Fig 1.
The experimental procedure of our clustering algorithm comparison.
Fig 2.
The impact of μ on cluster detectability, visualized using the spring-embedded “ForceAtlas” algorithm of the software Gephi [12].
Fig 3.
A sample network for which modularity ≈ 0.34, conductance ≈ 0.55, and coverage = 0.75.
The color of each node defines its cluster.
Table 1.
LFR benchmark graph parameters.
Fig 4.
A matrix of violin plots illustrating the synthetic graph experiment results at μ = 0.40.
We drew each “violin” using a Gaussian kernel density estimation. Red lines indicate the minimum, maximum, and mean of the data.
Fig 5.
A matrix of violin plots illustrating the synthetic graph experiment results at μ = 0.50.
We drew each “violin” using a Gaussian kernel density estimation. Red lines indicate the minimum, maximum, and mean of the data.
Fig 6.
A matrix of violin plots illustrating the synthetic graph experiment results at μ = 0.60.
We drew each “violin” using a Gaussian kernel density estimation. Red lines indicate the minimum, maximum, and mean of the data.
Fig 7.
(A) A comparison of clustering algorithm performance by modularity on the real-world graphs.
(B) A comparison of clustering algorithm performance by conductance on the real-world graphs. (C) A comparison of clustering algorithm performance by coverage on the real-world graphs.