geom_lmboot: Do a balanced bootstrap on a linear model.

In JackAHutchings/jahrfun: JAH Miscellaneous R Functions

Do a balanced bootstrap on a linear model.

Description

This function calculates a linear regression of y versus x. The data are bootstrapped to generate confidence intervals around the slope and intercept. Optionally, the r2 value may also be calculated. By default, the geom prints the text of the results in the upper-left corner of the plot and also plots the fit line. The bootstrapped prediction interval around the fit may also be plotted.

Usage

geom_lmboot(
  mapping = NULL,
  data = NULL,
  stat = "identity",
  show.legend = NA,
  position = "identity",
  ...,
  n = 10000,
  show.fit = T,
  show.label = T,
  linesize = 1,
  location = "tl",
  method = "ols",
  m = T,
  m.95 = T,
  int = T,
  int.95 = T,
  r2 = T,
  r2.95 = F,
  p = T,
  fontfamily = "serif",
  ci.width = 95,
  show.pred.band = F,
  fill = "blue",
  pred.alpha = 0.25,
  pred.limits = c(NA, NA),
  pred.steps = 25,
  fontsize = 6,
  coef.digits = 2,
  nudge_x = 0,
  nudge_y = 0,
  inherit.aes = T
)

Arguments

mapping	Set of aesthetic mappings created by aes or aes_.
data	The data to be displayed in this layer.
stat	The statistical transformation to use on the data for this layer, as a string.
n	Bootstrap N, defaults to 10000
show.fit	Should the regression fit line be shown? Boolean, defaults to TRUE.
show.label	Should the regression fit summary be shown? Boolean, defaults to TRUE.
linesize	The size of the fit line (in points).
location	Location to print fit results. Defaults to 'tl' (top-left), but may be 'tr', 'bl', or 'br'.
method	Regression model of the slope. Defaults to 'ols' for Ordinary Least Squares, but may also be 'rma' to perform a Reduced Major Axis (i.e., Standard Major Axis - not Ranged)
m	Should the slope coefficient be shown? Boolean, defaults to TRUE.
m.95	Should the slope's confidence interval be shown? Boolean, defaults to TRUE.
int	Should the intercept be shown? Boolean, defaults to TRUE.
int.95	Should the intercept's confidence interval be shown? Boolean, defaults to TRUE.
r2	Should Pearson's r-squared be shown? Boolean, defaults to TRUE.
r2.95	Should the r-squared confidence interval be shown? Boolean, defaults to FALSE. This can be misleading for poor fits because the calculated r value may be either negative or positive and is then squared. Use with caution.
p	Should the slope's p-value be printed? This only prints if the hypothetical 'p' is above or below the significance level chosen. I.e., does the slope's confidence interval overlap with zero?
fontfamily	Font family as a string to use for label plotting.
ci.width	The width of the confidence interval to use (in percents).
show.pred.band	Should the prediction interval be calculated and shown? Boolean. This is computationally intensive because it is also bootstrapped.
fill	Color of the prediction interval.
pred.alpha	Alpha of the prediction interval.
pred.steps	Number of steps to calculate the prediction interval at. Increase for higher resolution at cost of computation time.
fontsize	Font size as an integer (in pointS). Used for lable plotting.
coef.digits	Number of digits to round to in the fit label.
nudge_x	Manual X-axis adjustment of the text. In plotting units (where plotting area ranges from 0 to 1).
nudge_y	Manual Y-axis adjustment of the text. In plotting units (where plotting area ranges from 0 to 1).

Aesthetics

The geom understands the following aesthetics (required are in bold):

• x

• y

• xsd (Vector with length 1 or equal to X. If present, used as the SD term of an rnorm call for each X value during bootstrapping)

• ysd (Vector with length 1 or equal to Y. If present, used as the SD term of an rnorm call for each Y value during bootstrapping)

JackAHutchings/jahrfun documentation built on June 8, 2025, 3:09 a.m.