Skip to contents

Plot DAG (Directed Acyclic Graph) from exametrika

Source: R/plotGraph_gg.R
plotGraph_gg.Rd

Creates a ggplot2-based visualization of DAG structures from exametrika Bayesian network models. Supports BNM, LDLRA, LDB, and BINET models. Design based on TDE figures with items as rounded rectangles and fields as diamonds.

Usage

plotGraph_gg(
  data,
  layout = "sugiyama",
  direction = "TB",
  node_size = 12,
  label_size = 4,
  arrow_size = 3,
  title = TRUE,
  colors = NULL,
  fill_by = c("crr", "type"),
  fill_limits = NULL,
  show_isolated = FALSE,
  show_legend = NULL,
  legend_position = "right"
)

Arguments

data

An object of class "exametrika" with model type BNM, LDLRA, LDB, or BINET.

layout

Character string specifying the layout algorithm. Options include "sugiyama" (default, hierarchical), "fr" (Fruchterman-Reingold), "kk" (Kamada-Kawai), "tree", "stress".

direction

Character string for hierarchical layout direction. Controls the direction of arrows and node placement in hierarchical layouts. Only applies when layout = "sugiyama" (default).

  • "TB" (default): Top-to-Bottom. Source (parent) nodes at top, arrows point downward. Traditional hierarchical view.

  • "BT": Bottom-to-Top. Source nodes at bottom, arrows point upward. Use when representing progression or growth (e.g., learning paths). This was the default before version 1.3.0.

  • "LR": Left-to-Right. Source nodes at left, arrows point right. Good for wide displays or process flows.

  • "RL": Right-to-Left. Source nodes at right, arrows point left. Alternative horizontal layout.

For other layouts (e.g., "fr", "kk"), this parameter is ignored.

node_size

Numeric value for node size. Default is 12.

label_size

Numeric value for label text size inside nodes. Default is 4.

arrow_size

Numeric value for arrow size. Default is 3.

title

Logical or character. If TRUE (default), display an auto-generated title. If FALSE, no title. If a character string, use it as a custom title. For LDLRA, a single string gets " - Rank N" appended; a character vector of length equal to the number of ranks assigns each element as the title for the corresponding rank.

colors

Character vector. Colors for node types. For BNM with fill_by = "crr": two colors for the low and high ends of the correct-response-rate gradient. For BNM with fill_by = "type" and LDLRA: single color for Item nodes. For LDB: single color for Field nodes. For BINET: two colors for Class and Field nodes (positional or named). If NULL (default), uses the default node type colors (Item=purple, Field=green, Class=blue).

fill_by

Character. BNM only: how Item nodes are filled. "crr" (default) shades each node by the item's correct response rate (data$crr), so easy and hard items can be told apart at a glance; "type" uses a single color for all Item nodes (the behavior before version 1.3.0). Ignored for LDLRA, LDB and BINET.

fill_limits

Numeric vector of length 2, or NULL (default). Range of the correct-response-rate color scale. NULL stretches the scale over the observed rates, which keeps differences visible when all rates are close (e.g. rare symptoms); c(0, 1) fixes the scale so plots from different data sets can be compared. Used only with fill_by = "crr".

show_isolated

Logical. Whether to draw nodes with no edges (items or fields unrelated to all others) for BNM, LDLRA and LDB. FALSE (default) leaves them out of the graph but lists them in a caption under the plot (e.g. "Isolated items: 5, 9"), so it stays visible that they exist and are unconnected. TRUE draws them in a row of their own after the last layer of the hierarchical layout. Ignored for BINET.

show_legend

Logical. If TRUE, display the legend (the correct-response-rate color bar for BNM with fill_by = "crr", otherwise the node type legend). If NULL (default), the color bar is shown for BNM with fill_by = "crr" and no legend otherwise.

legend_position

Character. Position of the legend. One of "right" (default), "top", "bottom", "left", "none".

Value

A list of ggplot objects. For BNM, a single-element list. For LDLRA/LDB, one plot per rank/class. For BINET, the integrated graph.

Details

Design follows TDE figure specifications:

  • BNM: Items shown as circles with numbers inside, shaded by correct response rate (or a single purple with fill_by = "type")

  • LDLRA: Items as circles per rank/class, one plot per rank

  • LDB: Fields shown as green diamonds, one plot per rank

  • BINET: Integrated graph with Class nodes (blue squares) and Field nodes (green diamonds) as intermediates. Field nodes are smaller and placed between the Class nodes they connect

Direction Guidelines:

Choose the direction that best represents your data structure:

  • Use TB (default) for traditional hierarchies or causal flows

  • Use BT for learning progression, skill development, or upward growth

  • Use LR/RL for temporal sequences or wide-format displays

Auto-Scaling:

The function automatically scales node and arrow sizes based on graph complexity:

  • 1-5 nodes: 100\

  • 6-10 nodes: 80\

  • 11-20 nodes: 60\

  • 21+ nodes: 50\

Arrows maintain a minimum size of 2mm to ensure visibility.

Note

The linetype common option is not applicable to DAG visualization, as edges are drawn as directed arrows rather than standard lines.

Examples

library(exametrika)
library(ggExametrika)

# BNM example
DAG <- matrix(c(
  "Item01", "Item02", "Item02", "Item03",
  "Item02", "Item04", "Item03", "Item05",
  "Item04", "Item05"
), ncol = 2, byrow = TRUE)
items <- J5S10$ItemLabel
adj <- matrix(0, length(items), length(items), dimnames = list(items, items))
adj[DAG] <- 1
result <- BNM(J5S10, adj_matrix = adj)

# Default: Top-to-Bottom (parents on top, arrows point downward)
plots <- plotGraph_gg(result)
plots[[1]]


# Alternative directions
plotGraph_gg(result, direction = "BT")[[1]] # Bottom-to-Top

plotGraph_gg(result, direction = "LR")[[1]] # Left-to-Right

plotGraph_gg(result, direction = "RL")[[1]] # Right-to-Left


# Force-directed layout (no direction parameter)
plotGraph_gg(result, layout = "fr")[[1]]


# Fixed 0-1 color scale (comparable across plots) or a single color
plotGraph_gg(result, fill_limits = c(0, 1))[[1]]

plotGraph_gg(result, fill_by = "type")[[1]]


# Custom title
plotGraph_gg(result, title = "My Network Model")[[1]]