ggmid.mid.breakdown: Plot MID Breakdowns with ggplot2

In midr: Learning from Black-Box Models by Maximum Interpretation Decomposition

Plot MID Breakdowns with ggplot2

Description

For "mid.breakdown" objects, ggmid() visualizes the breakdown of a prediction by component functions.

Usage

## S3 method for class 'mid.breakdown'
ggmid(
  object,
  type = c("waterfall", "barplot", "dotchart"),
  theme = NULL,
  terms = NULL,
  max.nterms = 15L,
  width = NULL,
  vline = TRUE,
  catchall = "others",
  format = c("%t=%v", "%t"),
  ...
)

## S3 method for class 'mid.breakdown'
autoplot(object, ...)

Arguments

object	a "mid.breakdown" object to be visualized.
type	the plotting style. One of "waterfall", "barplot" or "dotchart".
theme	a character string or object defining the color theme. See color.theme for details.
terms	an optional character vector specifying which terms to display.
max.nterms	the maximum number of terms to display in the plot. Less important terms will be grouped into a "catchall" category.
width	a numeric value specifying the width of the bars.
vline	logical. If TRUE, a vertical line is drawn at the zero or intercept line.
catchall	a character string for the catchall label.
format	a character string or character vector of length two to be used as the format of the axis labels. Use "%t" for the term name (e.g., "Wind") and "%v" for the values (e.g., "30").
...	optional parameters passed on to the main layer.

Details

This is an S3 method for the ggmid() generic that creates a breakdown plot from a "mid.breakdown" object, visualizing the contribution of each component function to a single prediction.

The type argument controls the visualization style. The default, type = "waterfall" (default), creates a waterfall plot that shows how the prediction is built up from the intercept, with each term's contribution sequentially added or subtracted. The type = "barplot" option creates a standard bar plot where the length of each bar represents the magnitude of the term's contribution. The type = "dotchart" option creates a dot plot showing the contribution of each term as a point connected to a zero baseline.

Value

ggmid.mid.breakdown() returns a "ggplot" object.

See Also

mid.breakdown, ggmid, plot.mid.breakdown

Examples

data(diamonds, package = "ggplot2")
set.seed(42)
idx <- sample(nrow(diamonds), 1e4)
mid <- interpret(price ~ (carat + cut + color + clarity)^2, diamonds[idx, ])
mbd <- mid.breakdown(mid, diamonds[1L, ])

# Create a waterfall plot
ggmid(mbd, type = "waterfall")

# Create a bar plot with a different theme
ggmid(mbd, type = "barplot", theme = "highlight")

# Create a dot chart
ggmid(mbd, type = "dotchart", size = 3)

midr documentation built on Sept. 11, 2025, 1:07 a.m.