Confusion Plot Configuration

Portable confusion counts with reference rows and predicted columns. Data consists of long records with class labels and a frequency, optionally grouped into panels. Missing class labels represent omitted pairs; absent class combinations have zero frequency. All panels use the same class order. Compiles to HeatmapSeriesOption in src/chart/heatmap/HeatmapSeries.ts, GridOption in src/coord/cartesian/GridModel.ts, and continuous visual maps. to_list() emits semantic config keys; compile() emits ECharts options. draw() fits square count cells with aligned marginal summaries to the available canvas. Missing-pair omissions are reported in the console. The cell counts show included observations without a separate sample-size label. Color fades and marginal backgrounds follow the active theme unless explicitly overridden.

Usage

ConfusionConfig(
  dat_path = NULL,
  title = NULL,
  origin = NULL,
  writer = NULL,
  reference = "reference",
  predicted = "predicted",
  count = "n",
  panel = NULL,
  classes = NULL,
  show_metrics = TRUE,
  digits = 2L,
  ncol = 2L,
  correct_color = "#0F6A66",
  incorrect_color = "#BE2E5F",
  low_color = NULL,
  summary_color = NULL,
  font_size = 12,
  xlab = "Predicted",
  ylab = "Reference"
)

Arguments

dat_path

Optional Character: Path to the data, read at draw time. The serializable alternative to passing data to draw().

title

Optional Character: Chart title. Declared here because it is the one property every chart type has; palette is not (four take a colormap instead), and axis labels and margins are cartesian-only.

origin

Optional Named character {"user", "default", "derived"}: Where each value came from, one entry per settable property. Absent on an authored config; written by the interface that resolved it.

writer

Optional Named character: Which interface wrote the config, as name and version. Absent on an authored config.

reference

Character: Column containing reference class labels.

predicted

Character: Column containing predicted class labels.

count

Character: Column containing nonnegative integer frequencies.

panel

Optional Character: Optional panel-label column; unset draws one matrix.

classes

Optional Character vector: Ordered class labels shared by every panel; unset preserves first appearance.

show_metrics

Logical: Show per-class rates, accuracy, and balanced accuracy.

digits

Integer: Decimal places for rates in cells and hover text.

ncol

Integer: Maximum number of panels per row.

correct_color

Character: Six-digit hex color at unit fraction for correct predictions.

incorrect_color

Character: Six-digit hex color at unit fraction for incorrect predictions.

low_color

Optional Character: Six-digit hex color at zero row fraction; unset uses the active theme background.

summary_color

Optional Character: Six-digit hex background for metric cells; unset uses a faint neutral tint of the active theme background.

font_size

Numeric: Preferred cell-label font size in pixels. Constrained surfaces reduce text to fit, down to 8 pixels. Long class labels wrap and column labels rotate when needed. Increase figure dimensions if labels and metrics cannot fit without overlap.

xlab

Character: Predicted-class axis label.

ylab

Character: Reference-class axis label.

Value

A ConfusionConfig object.

Statistical semantics

Color intensity is the fraction within each reference row; labels show raw counts. Sensitivity, specificity, PPV, and NPV are one-versus-rest rates for each named class. Balanced accuracy is mean class recall, including every declared class. Zero denominators produce NA, never zero; consequently balanced accuracy is NA if any declared class has no reference observations. Summaries of pooled resamples describe pooled predictions, not mean fold performance. Repeated observations are counted each time they appear.

Examples

cfg <- setup_ConfusionConfig()
draw(cfg, data = data.frame(reference = c("yes", "no"),
  predicted = c("yes", "no"), n = c(8, 12)))