Plot IC Acceptance Matrix with Optional Rate-of-Improvement Overlay

Create a two-layer base R plot that visualizes information criterion (IC) scores across a sequence of sub-model evaluations, highlighting which steps were accepted vs rejected. Optionally, a secondary y-axis overlays the rate of improvement (first difference of IC scores) as a line with markers.

Usage

plot_ic_acceptance_matrix(
  matrix_data,
  plot_title = "IC Acceptance Matrix Scatter Plot",
  plot_rate_of_improvement = TRUE
)

Arguments

matrix_data: A two-column matrix or data.frame. Column 1 must be numeric IC scores in evaluation order; Column 2 must be a logical or numeric flag (0/1) indicating whether the step was accepted.
plot_title: character(1). Title to draw above the plot.
plot_rate_of_improvement: logical(1). If TRUE, overlay the first differences of the IC series on a secondary (right) y-axis along with a horizontal reference line at zero.

Value

Invisibly returns NULL. Called for its plotting side effects.

Details

The function expects a two-column object where:

Column 1 contains the IC score at each step (numeric; lower is better).
Column 2 contains an indicator for acceptance (0 = rejected, 1 = accepted).

The first IC value is treated as the baseline and is plotted as a larger black point and labeled. Accepted steps are drawn as blue filled points connected by a thin line; rejected steps are drawn as small red crosses. When plot_rate_of_improvement = TRUE, the function overlays a secondary y-axis on the right that shows diff(IC) values (the per-step change in IC; more negative implies improvement).

The function uses only base graphics. It sets plot margins and mgp via par(), and (when overlaying) uses par(new = TRUE) to layer the IC plot over the rate-of-improvement axes. It does not restore previous graphical parameters.

Axes and scaling. Tick marks for the primary (IC) x/y axes are computed with pretty() to give clean bounds. The secondary axis for the rate of improvement uses fixed limits (c(-400, 150)) inside the function; adjust in source if your expected diff(IC) range differs substantially.

#' @examples ic <- c(-1000, -1012, -1008, -1025, -1020, -1030) accepted <- c(1, 0, 1, 0, 1) # steps 2..6 relative to baseline mat <- cbind(ic, c(1, accepted)) # mark baseline as accepted for plotting plot_ic_acceptance_matrix(mat, plot_title = "IC Path")

Avoid non-ASCII glyphs in titles on CRAN/CI:

plot_ic_acceptance_matrix(mat, plot_rate_of_improvement = TRUE)