Plot a fitted CTA tree

Native base-R CTA visualization. Calls cta_plot_data for layout; uses only base graphics - no external package dependencies.

Split (internal) nodes are drawn as ellipses; terminal endpoint nodes are drawn as rectangles; edges are directed arrows. Split nodes show the split attribute, node-level ESS or WESS, and observation count. Without target_class, leaf nodes show the majority-class prediction and observation count. With target_class, leaf nodes show the target-class count, percentage, predicted class, and stage from cta_staging_table. Edge labels show the branch condition (e.g. "V14<=0.5").

Color note: when target_class is supplied, endpoint fill colors are assigned by ascending rank of each endpoint's target-class proportion within this tree. Colors encode relative position in the endpoint distribution and do not imply clinical thresholds or categories. Supply a custom palette via endpoint_palette to change the color encoding. Use show_caption = TRUE to render an explicit note on the plot.

cta_plot_data is the renderer-independent data contract. This function (plot.cta_tree) is the current native base-R renderer.

Usage

# S3 method for class 'cta_tree'
plot(x,
     target_class = NULL, class_labels = NULL, digits = 1,
     main = "CTA Tree", show_counts = TRUE, show_stage = TRUE,
     endpoint_palette = NULL, endpoint_fill = "#D9F7E6",
     split_fill = "#D9EAF7", node_col_split = NULL,
     node_col_leaf = NULL, edge_col = "grey40",
     border_col = "grey30", text_col = "black",
     arrow_col = NULL, show_caption = FALSE,
     cex = 0.75, ...)

Arguments

x: A cta_tree from oda_cta_fit.
target_class: Integer target class for endpoint annotation; passed to cta_plot_data. NULL (default) produces a structural plot without endpoint enrichment.
class_labels: Optional display names for class labels; passed to cta_plot_data. See cta_plot_data for accepted formats.
digits: Decimal places for percentage labels in enriched endpoint nodes; passed to cta_plot_data. Default 1.
main: Character plot title. Default "CTA Tree".
show_counts: Logical; include n=a/b raw counts in endpoint labels when target_class is supplied. Default TRUE.
show_stage: Logical; include Stage s line in endpoint labels when target_class is supplied. Default TRUE.
endpoint_palette: Palette for endpoint fill colors when target_class is supplied; passed to cta_plot_data. NULL uses the default gradient. Accepts a palette function(n) or a character vector of colors.
endpoint_fill: Default fill colour for leaf (terminal) nodes when target_class is NULL. Default "#D9F7E6" (light green).
split_fill: Fill colour for split (internal) ellipse nodes. Default "#D9EAF7" (light blue).
node_col_split: Legacy alias for split_fill; overrides it when non-NULL.
node_col_leaf: Legacy alias for endpoint_fill; overrides it when non-NULL.
edge_col: Colour for directed edge arrows. Default "grey40".
border_col: Border colour for all nodes. Default "grey30".
text_col: Text colour for node labels. Default "black".
arrow_col: Arrow colour for directed edges. NULL (default) uses edge_col.
show_caption: Logical; if TRUE and target_class is supplied, adds a bottom caption: "Endpoint fill: relative target-class proportion within this tree. Not a clinical threshold." Default FALSE.
cex: Text expansion factor for node labels. Default 0.75.
...: Unused; included for S3 compatibility.

Value

invisible(pd), where pd is the cta_plot_data list used to render the plot. The caller can inspect layout coordinates, enrichment columns, and endpoint annotations from the returned object.

Examples

data(mtcars)
X    <- mtcars[, c("cyl", "disp", "hp", "wt")]
y    <- as.integer(mtcars$am)
tree <- suppressMessages(
  oda_cta_fit(X, y, mindenom = 5L, mc_iter = 500L, mc_seed = 42L,
              loo = "off")
)

# Structural plot
plot(tree)


# Target-class enriched plot with custom labels
plot(tree, target_class = 1L,
     class_labels = c("0" = "Manual", "1" = "Auto"))


# Custom palette (white to dark red)
plot(tree, target_class = 1L,
     endpoint_palette = c("#ffffff", "#c62828"))

Usage

Arguments

Value

See also

Examples