nmfkc.net.DOT: Generate a Graphviz DOT Diagram for a Symmetric NMF Network

In nmfkc: Non-Negative Matrix Factorization with Kernel Covariates

Generate a Graphviz DOT Diagram for a Symmetric NMF Network

Description

Creates a Graphviz DOT script that visualizes the two-layer structure of a symmetric NMF model (Y \approx X C X^\top).

The resulting diagram displays:

• outer nodes: the original network nodes (rows/columns of Y),

• inner nodes: the latent basis/group nodes (columns of X),

• directed edges from basis to node: membership weights from X,

• undirected edges between basis nodes: inter-group interactions from C matrix (for tri-symmetric models only).

Usage

nmfkc.net.DOT(
  result,
  threshold = 0.01,
  sig.level = 0.1,
  weight_scale = 5,
  weight_scale_xy = 1,
  weight_scale_xx = weight_scale,
  rankdir = "TB",
  fill = TRUE,
  hide.isolated = TRUE,
  Y.label = NULL,
  X.label = NULL,
  Y.title = "Nodes (Y)",
  X.title = "Basis (X)",
  show.theta = NULL,
  signed = inherits(result, "nmfkc.net.signed"),
  cluster.box = c("none", "normal", "faint", "invisible"),
  layout = c("neato", "fdp", "twopi", "circo", "dot"),
  X.color = NULL,
  Y.cluster = c("soft", "hard")
)

Arguments

result	A list returned by nmfkc() with Y.symmetric = "bi" or "tri", or the newer nmfkc.net with type = "bi" or "tri". If inference results are present (from nmfkc.net.inference), C edges are decorated with significance stars.
threshold	Minimum coefficient value to display an edge.
sig.level	Significance level for filtering C edges (if inference results are present). Set to NULL to show all edges above threshold.
weight_scale	Base scaling factor for edge widths.
weight_scale_xy	Scaling factor for X edges (basis -> node).
weight_scale_xx	Scaling factor for C edges (basis <-> basis).
rankdir	Graphviz rank direction ("TB" or "LR").
fill	Logical; whether nodes are filled with color.
hide.isolated	Logical; if TRUE, omit outer nodes with no X edge above threshold.
Y.label	Character vector of labels for outer nodes.
X.label	Character vector of labels for basis nodes.
Y.title	Cluster title for outer nodes.
X.title	Cluster title for basis nodes.
show.theta	Logical or NULL. Whether to draw C edges between basis nodes. NULL = auto-detect (TRUE for tri, FALSE for bi).
signed	Logical. If TRUE, C is treated as a signed matrix: positive entries rendered as solid edges and negative as dashed, with edge visibility threshold on |C|. Default is inherits(result, "nmfkc.net.signed") so that results from nmfkc.net are auto-detected.
cluster.box	Style of cluster box: "none", "normal", "faint", "invisible".
layout	Graphviz layout engine, in recommended order: "neato" (default; spring model, clearest for small/medium community graphs), "fdp" (force-directed, scales to larger graphs), "twopi" (radial), "circo" (circular), "dot" (hierarchical). For community networks, "neato" or "fdp" with a raised threshold (e.g.\ 0.2–0.3) separate the groups best.
X.color	Color palette for basis nodes (length Q).
Y.cluster	Coloring mode for outer nodes: "soft" (weighted mix) or "hard" (most probable basis color).

Value

A character string of class c("nmfkc.net.DOT", "nmfkc.DOT") representing a valid Graphviz DOT script. Use plot() to render (requires DiagrammeR).

Lifecycle

This function is experimental. The interface may change in future versions; details are to be described in an upcoming paper.

See Also

nmfkc.net.inference, nmfkc.DOT, plot.nmfkc.DOT

Examples

library(nmfkc)
Y <- matrix(c(0,1,1,0,0,0,
              1,0,1,0,0,0,
              1,1,0,1,0,0,
              0,0,1,0,1,1,
              0,0,0,1,0,1,
              0,0,0,1,1,0), 6, 6)
res <- nmfkc.net(Y, rank = 2, type = "tri", nstart = 20)
dot <- nmfkc.net.DOT(res)
plot(dot)

nmfkc documentation built on July 14, 2026, 1:07 a.m.