Residuals from model fit as segments

stat_fit_deviations fits a linear model and returns fitted values and residuals ready to be plotted as segments.

Usage

stat_fit_deviations(
  mapping = NULL,
  data = NULL,
  geom = "segment",
  position = "identity",
  ...,
  method = "lm",
  method.args = list(),
  n.min = 2L,
  formula = NULL,
  na.rm = FALSE,
  orientation = NA,
  show.legend = FALSE,
  inherit.aes = TRUE
)

stat_fit_fitted(
  mapping = NULL,
  data = NULL,
  geom = "point",
  method = "lm",
  method.args = list(),
  n.min = 2L,
  formula = NULL,
  position = "identity",
  na.rm = FALSE,
  orientation = NA,
  show.legend = FALSE,
  inherit.aes = TRUE,
  ...
)

Arguments

mapping: The aesthetic mapping, usually constructed with aes. Only needs to be set at the layer level if you are overriding the plot defaults.
data: A layer specific dataset - only needed if you want to override the plot defaults.
geom: The geometric object to use display the data
position: The position adjustment to use for overlapping points on this layer
...: other arguments passed on to layer. This can include aesthetics whose values you want to set, not map. See layer for more details.
method: function or character If character, "lm", "rlm", "lqs", "rq" and the name of a function to be matched, possibly followed by the fit function's method argument separated by a colon (e.g. "rq:br"). Functions implementing methods must accept arguments to parameters formula, data, weights and method. A fitted() method must exist for the returned model fit object class.
method.args: named list with additional arguments.
n.min: integer Minimum number of distinct values in the explanatory variable (on the rhs of formula) for fitting to the attempted.
formula: a "formula" object. Using aesthetic names instead of original variable names.
na.rm: a logical indicating whether NA values should be stripped before the computation proceeds.
orientation: character Either "x" or "y" controlling the default for formula.
show.legend: logical. Should this layer be included in the legends? NA, the default, includes if any aesthetics are mapped. FALSE never includes, and TRUE always includes.
inherit.aes: If FALSE, overrides the default aesthetics, rather than combining with them. This is most useful for helper functions that define both data and aesthetics and should not inherit behaviour from the default plot specification, e.g. borders.

Details

This stat can be used to automatically highlight residuals as segments in a plot of a fitted model equation. This stat only generates the residuals, the predicted values need to be separately added to the plot, so to make sure that the same model formula is used in all steps it is best to save the formula as an object and supply this object as argument to the different statistics.

A ggplot statistic receives as data a data frame that is not the one passed as argument by the user, but instead a data frame with the variables mapped to aesthetics. In other words, it respects the grammar of graphics and consequently within the model formula names of aesthetics like $x$ and $y$ should be used instead of the original variable names. This helps ensure that the model is fitted to the same data as plotted in other layers.

Note

In the case of method = "rq" quantiles are fixed at tau = 0.5 unless method.args has length > 0. Parameter orientation is redundant as it only affects the default for formula but is included for consistency with ggplot2.

Computed variables

Data frame with same nrow as data as subset for each group containing five numeric variables.

x: x coordinates of observations
x.fitted: x coordinates of fitted values
y: y coordinates of observations
y.fitted: y coordinates of fitted values

To explore the values returned by this statistic we suggest the use of geom_debug. An example is shown below, where one can also see in addition to the computed values the default mapping of the fitted values to aesthetics xend and yend.

Examples

# generate artificial data
library(MASS)

set.seed(4321)
x <- 1:100
y <- (x + x^2 + x^3) + rnorm(length(x), mean = 0, sd = mean(x^3) / 4)
my.data <- data.frame(x, y)

# plot residuals from linear model
ggplot(my.data, aes(x, y)) +
  geom_smooth(method = "lm", formula = y ~ x) +
  stat_fit_deviations(method = "lm", formula = y ~ x, colour = "red") +
  geom_point()


# plot residuals from linear model with y as explanatory variable
ggplot(my.data, aes(x, y)) +
  geom_smooth(method = "lm", formula = y ~ x, orientation = "y") +
  stat_fit_deviations(method = "lm", formula = x ~ y, colour = "red") +
  geom_point()


# as above using orientation
ggplot(my.data, aes(x, y)) +
  geom_smooth(method = "lm", orientation = "y") +
  stat_fit_deviations(orientation = "y", colour = "red") +
  geom_point()
#> `geom_smooth()` using formula = 'y ~ x'


# both regressions and their deviations
ggplot(my.data, aes(x, y)) +
  geom_smooth(method = "lm") +
  stat_fit_deviations(colour = "blue") +
  geom_smooth(method = "lm", orientation = "y", colour = "red") +
  stat_fit_deviations(orientation = "y", colour = "red") +
  geom_point()
#> `geom_smooth()` using formula = 'y ~ x'
#> `geom_smooth()` using formula = 'y ~ x'


# give a name to a formula
my.formula <- y ~ poly(x, 3, raw = TRUE)

# plot linear regression
ggplot(my.data, aes(x, y)) +
  geom_smooth(method = "lm", formula = my.formula) +
  stat_fit_deviations(formula = my.formula, colour = "red") +
  geom_point()


ggplot(my.data, aes(x, y)) +
  geom_smooth(method = "lm", formula = my.formula) +
  stat_fit_deviations(formula = my.formula, method = stats::lm, colour = "red") +
  geom_point()


# plot robust regression
ggplot(my.data, aes(x, y)) +
  stat_smooth(method = "rlm", formula = my.formula) +
  stat_fit_deviations(formula = my.formula, method = "rlm", colour = "red") +
  geom_point()


# plot robust regression with weights indicated by colour
my.data.outlier <- my.data
my.data.outlier[6, "y"] <- my.data.outlier[6, "y"] * 10
ggplot(my.data.outlier, aes(x, y)) +
  stat_smooth(method = MASS::rlm, formula = my.formula) +
  stat_fit_deviations(formula = my.formula, method = "rlm",
                      mapping = aes(colour = after_stat(weights)),
                      show.legend = TRUE) +
  scale_color_gradient(low = "red", high = "blue", limits = c(0, 1),
                       guide = "colourbar") +
  geom_point()


# plot quantile regression (= median regression)
ggplot(my.data, aes(x, y)) +
  stat_quantile(formula = my.formula, quantiles = 0.5) +
  stat_fit_deviations(formula = my.formula, method = "rq", colour = "red") +
  geom_point()


# plot quantile regression (= "quartile" regression)
ggplot(my.data, aes(x, y)) +
  stat_quantile(formula = my.formula, quantiles = 0.75) +
  stat_fit_deviations(formula = my.formula, colour = "red",
                      method = "rq", method.args = list(tau = 0.75)) +
  geom_point()


# inspecting the returned data with geom_debug()
gginnards.installed <- requireNamespace("gginnards", quietly = TRUE)

if (gginnards.installed)
  library(gginnards)

# plot, using geom_debug() to explore the after_stat data
if (gginnards.installed)
  ggplot(my.data, aes(x, y)) +
    geom_smooth(method = "lm", formula = my.formula) +
    stat_fit_deviations(formula = my.formula, geom = "debug") +
    geom_point()

#> [1] "PANEL 1; group(s) -1; 'draw_function()' input 'data' (head):"
#>   xend       yend x          y x.fitted   y.fitted weights hjust PANEL group
#> 1    1 -3691.7638 1 -27205.450        1 -3691.7638       1     0     1    -1
#> 2    2 -2578.9186 2 -14242.651        2 -2578.9186       1     0     1    -1
#> 3    3 -1498.5419 3  45790.918        3 -1498.5419       1     0     1    -1
#> 4    4  -443.8109 4  53731.420        4  -443.8109       1     0     1    -1
#> 5    5   592.0973 5  -8028.578        5   592.0973       1     0     1    -1
#> 6    6  1616.0056 6 102863.943        6  1616.0056       1     0     1    -1
#>   orientation
#> 1          NA
#> 2          NA
#> 3          NA
#> 4          NA
#> 5          NA
#> 6          NA

if (gginnards.installed)
  ggplot(my.data.outlier, aes(x, y)) +
    stat_smooth(method = MASS::rlm, formula = my.formula) +
    stat_fit_deviations(formula = my.formula, method = "rlm", geom = "debug") +
    geom_point()

#> [1] "PANEL 1; group(s) -1; 'draw_function()' input 'data' (head):"
#>   xend       yend x           y x.fitted   y.fitted    weights hjust PANEL
#> 1    1 -2515.4775 1  -27205.450        1 -2515.4775 1.00000000     0     1
#> 2    2 -1531.5327 2  -14242.651        2 -1531.5327 1.00000000     0     1
#> 3    3  -573.5792 3   45790.918        3  -573.5792 1.00000000     0     1
#> 4    4   365.1016 4   53731.420        4   365.1016 1.00000000     0     1
#> 5    5  1291.2282 5   -8028.578        5  1291.2282 1.00000000     0     1
#> 6    6  2211.5193 6 1028639.429        6  2211.5193 0.07650433     0     1
#>   group orientation
#> 1    -1          NA
#> 2    -1          NA
#> 3    -1          NA
#> 4    -1          NA
#> 5    -1          NA
#> 6    -1          NA