cograph

mat <- matrix(c(
  0.0, 0.4, 0.2,
  0.3, 0.0, 0.3,
  0.2, 0.3, 0.0
), 3, 3, byrow = TRUE)
rownames(mat) <- colnames(mat) <- c("A", "B", "C")

# Any function accepts matrices directly
splot(mat)
centrality(mat)
communities(mat)

edges <- data.frame(
  from = c("A", "A", "B", "C"),
  to   = c("B", "C", "C", "A"),
  weight = c(0.5, 0.3, 0.8, 0.2)
)

splot(edges)
centrality(edges)

library(igraph)
g <- make_graph("Zachary")

splot(g)
centrality(g, measures = c("degree", "betweenness"))
communities(g, method = "louvain")

library(qgraph)
q <- qgraph(mat)

# Extracts layout, colors, shapes, edge styles — everything
from_qgraph(q)

library(igraph)
g <- make_graph("Zachary")

# igraph → matrix → cograph_network → back to igraph
mat <- to_matrix(g)
net <- as_cograph(mat)
g2  <- to_igraph(net)

# Keep only strong edges
splot(filter_edges(mat, weight > 0.3))

# Pipe-friendly
mat |>
  filter_edges(weight > 0.3) |>
  splot()

# Keep high-degree nodes
mat |>
  filter_nodes(degree >= 3) |>
  splot()

# Combine edge and node filters
mat |>
  filter_edges(weight > 0.3) |>
  filter_nodes(degree >= 2) |>
  splot()

# Top 5 nodes by PageRank
mat |>
  select_nodes(top = 5, by = "pagerank") |>
  splot()

# Top 3 by betweenness, expand to neighbors
mat |>
  select_nodes(top = 3, by = "betweenness") |>
  splot()

# Edges within the same community
mat |>
  select_edges(same_community) |>
  splot()

# Bridge edges (connect different components)
mat |>
  select_bridges() |>
  splot()

# 2-hop neighborhood of node "A"
mat |>
  select_neighbors(of = "A", order = 2) |>
  splot()

# All 25 measures at once
centrality(mat)

# Specific measures, sorted
centrality(mat,
  measures = c("degree", "betweenness", "pagerank"),
  sort_by = "pagerank",
  digits = 3
)

# Each returns a named numeric vector
centrality_degree(mat)
#> A B C
#> 4 3 5

centrality_pagerank(mat, damping = 0.9)
#>     A     B     C
#> 0.35  0.28  0.37

# Directional shorthands
centrality_instrength(mat)
centrality_outdegree(mat)

# Node size proportional to PageRank
splot(mat, scale_nodes_by = "pagerank")

# Directional shorthand + dampening
splot(mat,
  scale_nodes_by = "instrength",
  scale_nodes_scale = 0.5,
  node_size_range = c(3, 12)
)

# All edge measures
edge_centrality(mat)

# Find bridge edges
edge_centrality(mat, sort_by = "betweenness")

# Named vector shorthand
edge_betweenness(mat)

network_summary(mat)
#> ── Network Summary ──
#> Nodes: 8 | Edges: 24 | Directed: TRUE
#> Density: 0.429 | Reciprocity: 0.833
#> Transitivity: 0.612 | Diameter: 3
#> Components: 1 | Mean degree: 6.0

library(igraph)
g <- make_graph("Zachary")

# Via dispatcher
comm <- communities(g, method = "louvain")
print(comm)
#> Community detection: louvain
#> Communities: 4 | Modularity: 0.42
#> Sizes: 12  6  5  11

# Direct function
comm2 <- community_fast_greedy(g)

# Short alias
comm3 <- com_wt(g)  # walktrap

# Extract results
membership(comm)
n_communities(comm)
community_sizes(comm)
modularity(comm)

# Run Louvain 100 times, build consensus
comm <- community_consensus(g,
  method = "louvain",
  n_runs = 100,
  threshold = 0.5
)

# Compare two community structures
compare_communities(comm, comm2, method = "nmi")

# Automatic community coloring
comm <- community_louvain(g)
plot(comm)  # dispatches to splot with community colors

# Or manually with color_communities()
splot(g, node_fill = color_communities(comm))

# Mixed node shapes
splot(mat,
  node_shape = c("circle", "hexagon", "diamond", "square", "star"),
  node_fill = c("#E63946", "#457B9D", "#2A9D8F", "#E9C46A", "#F4A261"),
  layout = "circle"
)

# Donut charts (outer ring = proportion)
splot(mat,
  donut_fill = c(0.9, 0.7, 0.5, 0.3, 0.8),
  donut_color = "steelblue"
)

# Pie charts (inner segments)
splot(mat,
  node_shape = "pie",
  pie_values = list(c(0.5, 0.3, 0.2), c(0.4, 0.4, 0.2), ...),
  pie_colors = c("#E63946", "#457B9D", "#2A9D8F")
)

# Donut + Pie combo
splot(mat,
  donut_fill = c(0.95, 0.87, 0.72),
  donut_color = "#2E7D32",
  donut_show_value = TRUE,
  pie_values = prob_vals,
  pie_colors = c("#1976D2", "#FFA000", "#C62828")
)

splot(mat,
  curvature = 0.3,
  arrow_size = 0.5,
  edge_width = 3,
  edge_color = "#546E7A",
  edge_start_style = "dotted",
  edge_labels = TRUE,
  threshold = 0.05
)

splot(mat,
  label_size = 1.1,
  label_fontface = "bold",
  label_color = "#1A237E",
  label_position = "below"
)

plot_chord(mat,
  segment_colors = c("#E63946", "#457B9D", "#2A9D8F"),
  ticks = TRUE,
  tick_labels = TRUE
)

plot_heatmap(mat,
  show_values = TRUE,
  color_palette = "viridis",
  cluster_list = list(
    Group1 = c("A", "B"),
    Group2 = c("C", "D", "E")
  )
)

# From transition matrices
plot_transitions(list(mat1, mat2, mat3))

# Individual tracking from data frame
df <- data.frame(T1 = c("A","B","A"), T2 = c("B","A","C"), T3 = c("C","C","B"))
plot_trajectories(df)

# Dark theme for presentations
splot(mat, theme = "dark", layout = "circle")

# Register a custom theme
register_theme("my_theme", list(
  node_fill = "#264653",
  node_border_color = "white",
  edge_color = "#2a9d8f",
  background = "#0f172a"
))
splot(mat, theme = "my_theme")

# Register a custom layout
register_layout("my_layout", function(n, ...) {
  cbind(x = cos(seq(0, 2*pi, length.out = n+1)[1:n]),
        y = sin(seq(0, 2*pi, length.out = n+1)[1:n]))
})

mat |>
  cograph(layout = "circle") |>
  sn_nodes(
    fill = c("#E63946", "#457B9D", "#2A9D8F", "#E9C46A", "#F4A261"),
    shape = "hexagon",
    size = 5
  ) |>
  sn_edges(
    color = "weight",
    width = "weight",
    curvature = 0.3
  ) |>
  sn_theme("minimal") |>
  sn_palette("viridis") |>
  splot(title = "My Network")

mat |>
  cograph() |>
  sn_nodes(fill = "steelblue") |>
  sn_save("network.pdf", width = 8, height = 8)

# Or get a ggplot object for further customization
p <- mat |>
  cograph() |>
  sn_ggplot()
p + ggplot2::labs(title = "Custom title")

library(tna)
library(cograph)

fit <- tna(engagement)

# One line — layout, arrows, edge labels, donut fills all auto-configured
splot(fit)

# Customize on top of TNA defaults
splot(fit,
  layout = "circle",
  node_shape = "hexagon",
  theme = "dark",
  scale_nodes_by = "instrength"
)

# Group-level TNA models
gfit <- group_tna(engagement, group = engagement$Group)

# Plot each group
splot(gfit)

# Comparison heatmap across groups
plot_heatmap(gfit)

fit <- tna(engagement)

# Node size by in-strength (which states are transitioned TO most)
splot(fit, scale_nodes_by = "instrength")

# Combine with donut showing initial probabilities
splot(fit,
  donut_fill = fit$initial_probs,
  donut_color = "#1976D2",
  donut_show_value = TRUE,
  scale_nodes_by = "outstrength",
  layout = "circle"
)

library(tna)
fit <- tna(engagement)

# Bootstrap analysis
boot <- bootstrap(fit, iter = 1000)

# Dedicated S3 method — network with CI-based styling
splot(boot)

# Permutation test between two groups
perm <- permutation_test(gfit)

# Dedicated visualization
splot(perm)

# Or use the specialized plot function
plot_permutation(perm)

# Visualize state transitions as alluvial flows
plot_transitions(fit)

# Individual trajectory tracking
plot_trajectories(fit$data)