Skip to contents

Fit a Residual Plot of Your Linear Model

Source: R/lmr_plot.R
lmr.plot.Rd

lmr_plot() provides a visual diagnostic of the linearity assumption of the linear model. Provided a linear model fit by lm() in base R, the function extracts the model frame and creates a faceted scatterplot. For each facet, a linear smoother and LOESS smoother are estimated over the points. Users who run this function can assess just how much the linear smoother and LOESS smoother diverge. The more they diverge, the more the user can determine how much the linear model is a good fit as specified. The plot will also point to potential outliers that may need further consideration.

Usage

lmr_plot(
  mod,
  smoother = "loess",
  se = FALSE,
  span = 0.75,
  dummies = FALSE,
  warning = TRUE,
  degree = 2,
  robust = FALSE,
  ...
)

# S3 method for class 'lmr.plot'
print(x, ...)

Arguments

mod

a fitted model, ideally a simple linear model

smoother

defaults to "loess", and is passed to the 'method' argument for the non-linear smoother in geom_smooth().

se

logical, defaults to FALSE. If TRUE, gives standard error estimates with the assorted smoothers.

span

a numeric, defaults to .75. An adjustment to the smoother. Higher values permit smoother lines and might be warranted in the presence of sparse pockets of the data.

dummies

logical, defaults to FALSE. If FALSE, removes binary independent variables from the plot. If TRUE, facets appear for binary independent variables. You should probably keep this as FALSE for your own use cases.

warning

logical, defaults to TRUE. If TRUE, the plot suppresses assorted warnings from the LOESS smoother that would otherwise be cautioning you about things your eyes could otherwise see.

degree

takes either 1 or 2 and passes argument to geom_smooth() call in the function. 1 = linear local fits. 2 = quadratic local fits.

robust

logical, defaults to FALSE. If TRUE, employs robust smoothing in which large residuals are downweighted while fitting.

...

Additional arguments (passed to the scatterplot in lmr.plot(), not used in the print function

x

a ggplot object with this special 'lmr.plot' class

Value

lmr_plot() returns a faceted scatterplot as a ggplot2 object. The linear smoother is flat at 0 by definition and is depicted by a black dashed line. The LOESS smoother is a blue line. You can add cosmetic features to it after the fact. The function may spit warnings to you related to the LOESS smoother, depending your data and whether you have disabled the warnings in the function. I think these to be fine the extent to which this is really just a visual aid and an informal diagnostic for the linearity assumption in a specified model.

Details

This function makes an implicit assumption that there is no variable in the regression formula with the name ".resid".

It may be in your interest (for the sake of rudimentary diagnostic checks) to disable the standard error bands for particularly ill-fitting linear models.

This function will ignore interactions in your model, returning only the first-order terms in the model's specification.

Author

Steven V. Miller

Examples


set.seed(8675309) # Don't need this updating every time...
x1 <- rnorm(500)
x2 <- rnorm(500)
e <- rnorm(500)
y <- 5 + .5*x1 + .5*x2 + e

M1 <- lm(y ~ x1 + x2)

lmr_plot(M1)
#> `geom_smooth()` using formula = 'y ~ x'
#> `geom_smooth()` using formula = 'y ~ x'

lmr_plot(M1, color="black", pch=21)
#> `geom_smooth()` using formula = 'y ~ x'
#> `geom_smooth()` using formula = 'y ~ x'