EC-SBM community detection analysis: full pipeline and writeup

Implement community detection on 3 EC-SBM networks (polblogs, topology,
internet_as) using 5 methods (Leiden-Mod, Leiden-CPM at 0.1 and 0.01,
Infomap, graph-tool SBM). Compute AMI/ARI/NMI accuracy, cluster statistics,
and generate figures and LaTeX report.

13 files changed, 60 insertions, 0 deletions

diff --git a/results/figures/accuracy_table.tex b/results/figures/accuracy_table.tex
new file mode 100644
index 0000000..1f218cc
--- /dev/null
+++ b/results/figures/accuracy_table.tex
@@ -0,0 +1,28 @@
+\begin{table}[htbp]
+\centering
+\caption{Community detection accuracy (AMI, ARI, NMI) on EC-SBM networks.}
+\label{tab:accuracy}
+\begin{tabular}{llrrr}
+\toprule
+Network & Method & AMI & ARI & NMI \\
+\midrule
+polblogs & Leiden-Mod & 0.1226 & 0.0237 & 0.2575 \\
+ & Leiden-CPM(0.1) & 0.4267 & 0.1622 & 0.8151 \\
+ & Leiden-CPM(0.01) & 0.3151 & 0.0363 & 0.6869 \\
+ & Infomap & 0.1561 & 0.0290 & 0.3464 \\
+ & graph-tool SBM & 0.2426 & 0.0771 & 0.5432 \\
+\midrule
+topology & Leiden-Mod & 0.0855 & 0.0031 & 0.4146 \\
+ & Leiden-CPM(0.1) & 0.2401 & 0.0590 & 0.9366 \\
+ & Leiden-CPM(0.01) & 0.2339 & 0.0677 & 0.8724 \\
+ & Infomap & 0.1772 & 0.0338 & 0.7284 \\
+ & graph-tool SBM & 0.0949 & 0.0034 & 0.4741 \\
+\midrule
+internet_as & Leiden-Mod & 0.0946 & 0.0065 & 0.4773 \\
+ & Leiden-CPM(0.1) & 0.2335 & 0.0847 & 0.9350 \\
+ & Leiden-CPM(0.01) & 0.1873 & 0.0974 & 0.8354 \\
+ & Infomap & 0.1604 & 0.0438 & 0.7523 \\
+ & graph-tool SBM & 0.0705 & 0.0024 & 0.4196 \\
+\bottomrule
+\end{tabular}
+\end{table}
\ No newline at end of file
diff --git a/results/figures/cluster_sizes_internet_as.pdf b/results/figures/cluster_sizes_internet_as.pdf
new file mode 100644
index 0000000..4065559
--- /dev/null
+++ b/results/figures/cluster_sizes_internet_as.pdf
diff --git a/results/figures/cluster_sizes_polblogs.pdf b/results/figures/cluster_sizes_polblogs.pdf
new file mode 100644
index 0000000..74a80ad
--- /dev/null
+++ b/results/figures/cluster_sizes_polblogs.pdf
diff --git a/results/figures/cluster_sizes_topology.pdf b/results/figures/cluster_sizes_topology.pdf
new file mode 100644
index 0000000..1936f62
--- /dev/null
+++ b/results/figures/cluster_sizes_topology.pdf
diff --git a/results/figures/cluster_stats_table.tex b/results/figures/cluster_stats_table.tex
new file mode 100644
index 0000000..3955e78
--- /dev/null
+++ b/results/figures/cluster_stats_table.tex
@@ -0,0 +1,32 @@
+\begin{table}[htbp]
+\centering
+\caption{Cluster statistics summary for each network and method.}
+\label{tab:cluster_stats}
+\footnotesize
+\begin{tabular}{llrrrrrr}
+\toprule
+Network & Method & \#Clusters & Node Cov. & Mean Size & Mean Density & Mean Cond. & Mean Mix. \\
+\midrule
+polblogs & ground_truth & 70 & 0.483 & 8.4 & 0.767 & 0.783 & 0.905 \\
+ & Leiden-Mod & 9 & 1.000 & 136.0 & 0.648 & 0.388 & 0.098 \\
+ & Leiden-CPM(0.1) & 145 & 0.811 & 6.8 & 0.586 & 0.801 & 0.653 \\
+ & Leiden-CPM(0.01) & 45 & 0.700 & 19.0 & 0.795 & 0.644 & 0.388 \\
+ & Infomap & 23 & 1.000 & 53.2 & 0.571 & 0.522 & 0.131 \\
+ & graph-tool SBM & 26 & 1.000 & 47.1 & 0.270 & 0.929 & 0.891 \\
+\midrule
+topology & ground_truth & 1517 & 0.219 & 5.0 & 0.825 & 0.620 & 0.881 \\
+ & Leiden-Mod & 50 & 1.000 & 695.2 & 0.231 & 0.256 & 0.101 \\
+ & Leiden-CPM(0.1) & 4047 & 0.576 & 4.9 & 0.572 & 0.594 & 0.676 \\
+ & Leiden-CPM(0.01) & 1802 & 0.733 & 14.1 & 0.451 & 0.495 & 0.518 \\
+ & Infomap & 1274 & 1.000 & 27.3 & 0.352 & 0.458 & 0.261 \\
+ & graph-tool SBM & 93 & 1.000 & 373.8 & 0.225 & 0.760 & 0.679 \\
+\midrule
+internet_as & ground_truth & 1085 & 0.211 & 4.5 & 0.841 & 0.594 & 0.881 \\
+ & Leiden-Mod & 35 & 1.000 & 656.1 & 0.040 & 0.224 & 0.131 \\
+ & Leiden-CPM(0.1) & 2895 & 0.613 & 4.9 & 0.583 & 0.549 & 0.632 \\
+ & Leiden-CPM(0.01) & 1157 & 0.879 & 17.5 & 0.328 & 0.447 & 0.392 \\
+ & Infomap & 1052 & 1.000 & 21.8 & 0.320 & 0.427 & 0.262 \\
+ & graph-tool SBM & 43 & 1.000 & 534.0 & 0.160 & 0.687 & 0.693 \\
+\bottomrule
+\end{tabular}
+\end{table}
\ No newline at end of file
diff --git a/results/figures/edge_density_internet_as.pdf b/results/figures/edge_density_internet_as.pdf
new file mode 100644
index 0000000..fe6e985
--- /dev/null
+++ b/results/figures/edge_density_internet_as.pdf
diff --git a/results/figures/edge_density_polblogs.pdf b/results/figures/edge_density_polblogs.pdf
new file mode 100644
index 0000000..e4d502e
--- /dev/null
+++ b/results/figures/edge_density_polblogs.pdf
diff --git a/results/figures/edge_density_topology.pdf b/results/figures/edge_density_topology.pdf
new file mode 100644
index 0000000..c4458e5
--- /dev/null
+++ b/results/figures/edge_density_topology.pdf
diff --git a/results/figures/heatmap_ami.pdf b/results/figures/heatmap_ami.pdf
new file mode 100644
index 0000000..0ade898
--- /dev/null
+++ b/results/figures/heatmap_ami.pdf
diff --git a/results/figures/heatmap_ari.pdf b/results/figures/heatmap_ari.pdf
new file mode 100644
index 0000000..ef5182f
--- /dev/null
+++ b/results/figures/heatmap_ari.pdf
diff --git a/results/figures/heatmap_nmi.pdf b/results/figures/heatmap_nmi.pdf
new file mode 100644
index 0000000..bca5275
--- /dev/null
+++ b/results/figures/heatmap_nmi.pdf
diff --git a/results/figures/mixing_parameter.pdf b/results/figures/mixing_parameter.pdf
new file mode 100644
index 0000000..9cb57d0
--- /dev/null
+++ b/results/figures/mixing_parameter.pdf
diff --git a/results/figures/node_coverage.pdf b/results/figures/node_coverage.pdf
new file mode 100644
index 0000000..94e57fd
--- /dev/null
+++ b/results/figures/node_coverage.pdf