In vincentarelbundock/marginsplot: Plot (conditional) marginal effects using the 'margins' and 'ggplot2' packages

This package overrides plotting functions from the margins R package in order to produce ggplot2 objects. This makes it much easier for users to customize the look of their marginal effects and predicted probabilities plots.

To install this package, execute the following commands:

library(remotes)
install_github('vincentarelbundock/marginsplot')

To use the package, you need to load marginsplot after margins, to ensure that the new plotting functions take precedence. For example:

library(margins)
library(marginsplot)
mod <- lm(hp ~ wt*drat, data = mtcars)
cplot(mod, dx = 'drat', x = 'wt')

Alternatively, you could explicitly prioritize marginsplot using the conflicted package:

library(conflicted)
conflict_prefer('cplot', 'marginsplot')
conflict_prefer('plot', 'marginsplot')

The next section is a modified version of the margins README, which shows how to customize the output of marginsplot.

Modified margins README

Visualization

The package implements several useful additional features for summarizing model objects, including:

• A plot() method for the new "margins" class that ports Stata's marginsplot command.
• A plotting function cplot() to provide the commonly needed visual summaries of predictions or average marginal effects conditional on a covariate.
• A persp() method for "lm", "glm", and "loess" objects to provide three-dimensional representations of response surfaces or marginal effects over two covariates.
• An image() method for the same that produces flat, two-dimensional heatmap-style representations of persp()-type plots.

plot() and cplot() return ggplot2 objects, which can be further customized using themes and other ggplot2 options.

Marginal effects plots

Using the plot() method yields an aesthetically similar result to Stata's marginsplot:

library("webuse")
webuse::webuse("nhanes2")
mod3 <- glm(highbp ~ sex * agegrp * bmi, data = nhanes2, family = binomial)
summary(marg3 <- margins(mod3))
plot(marg3)

Conditional effects plots

In addition to the estimation procedures and plot() generic, margins offers several plotting methods for model objects. First, there is a new generic cplot() that displays predictions or marginal effects (from an "lm" or "glm" model) of a variable conditional across values of third variable (or itself). For example, here is a graph of predicted probabilities from a logit model:

mod4 <- glm(am ~ wt*drat, data = mtcars, family = binomial)
cplot(mod4, x = "wt", dx = 'drat')

And fitted values with a factor independent variable:

cplot(lm(Sepal.Length ~ Species, data = iris))

and a graph of the effect of drat across levels of wt:

cplot(mod4, x = "wt", dx = "drat", what = "effect")

Customizing the look of your plots

Since cplot() and plot() return ggplot2 objects, the look is very easy to customize. For example, we could flip the marginal effects plot on its side, change the colors and line type, rename the variables, and add a red line at 0:

library(ggplot2)
mod3 <- glm(highbp ~ sex * agegrp * bmi, data = nhanes2, family = binomial)
marg3 <- margins(mod3)
plot(marg3,
     level = .99,
     colour = 'turquoise', size = 1.5, linetype = 'dashed',
     term_map = c('Body Mass Index' = 'bmi','Gender' = 'sex')) + 
  geom_hline(yintercept = 0, colour = 'red') +
  xlab('The two variables I care about') +
  ylab('Average marginal effects with 95% confidence intervals') +
  coord_flip()

Notice that the agegrp does not appear in the plot above because it was omitted from the term_map argument. This gives you a convenient way to omit terms from the plot.

We can apply the same kinds of changes to graphs produced by cplot():

library(ggplot2)
cplot(mod4, x = "wt", dx = 'drat', what='effect', size=2, colour='blue', 
      fill = 'pink', alpha=1, linetype='dashed', rugplot=FALSE) + 
    theme_grey() +
    xlab('Custom horizontal label') +
    ylab('Custom vertical label') +
    ggtitle('This is a custom title')

You can even use one of many user-created themes that are floating around the web. For instance, here is a plot drawn with the ggpomological theme and the Homemade Apple font:

library(ggpomological)
cplot(mod4, x = "wt", dx = 'drat', what='effect', alpha=.5, fill='#efe1c6', colour='#6f5438') + 
    theme_pomological_fancy()

If you would like even more control over your plots, cplot() can also return a data frame of quantities of interest:

dat <- cplot(mod4, x = "wt", dx = "drat", what = "effect", draw = FALSE)
head(dat)

vincentarelbundock/marginsplot documentation built on May 30, 2020, 12:07 a.m.