A simple use case
First, let’s load the libraries we’ll use here.
library(ggplot2)
library(latex2exp)
library(equatiomatic)Next, we’ll need a model. Let’s fit a simple linear model to the palmerpenguins dataset (which is exported from {equatiomatic}) using bill length to predict body mass.
# Fit an lm model
m <- lm(body_mass_g ~ bill_length_mm, penguins)Now that we have our model, we can extract the equation using {equatiomatic}. We’ll do this showing the actual coefficients, since we’ll be using this to communicate the results in a figure.
extract_eq(m, use_coefs = TRUE)\[ \operatorname{\widehat{body\_mass\_g}} = 362.31 + 87.42(\operatorname{bill\_length\_mm}) \]
Great! we can also show the relationship between these variables using {ggplot2} as follows.
ggplot(penguins, aes(x = bill_length_mm, y = body_mass_g)) +
geom_point() +
geom_smooth(method = "lm")
This is all well and good, but how do we get the equation to be incorporated into the figure? First, let’s store the equation into an object. Importantly, we must also extract the equation with ital_vars = TRUE. Using this option will prevent the use of \operatorname, which cannot be parsed by {latex2exp}. Additionally, there are a few other things that {latex2exp} tends to have difficulty handling. In this case, for example, we have underscores in our variable names, and {latex2exp} doesn’t like that (let’s replace them with dashes). {latex2exp} also doesn’t like the display version of equations when there are escaped greek letters in the equation (double $). We’ll therefore convert the display equation to an inline equation (replacing the $$ with $).
# extract equation with `ital_vars = TRUE` to avoid the use of `\operatorname`
m_eq <- extract_eq(m, use_coef = TRUE, ital_vars = TRUE)
# swap escaped underscores for dashes
prep_eq <- gsub("\\\\_", "-", m_eq)
# swap display-style $$ with inline-style $
prep_eq <- paste("$", as.character(prep_eq), "$", sep = "")And now, we can use it with our figure directly. Below are a couple different examples.
ggplot(penguins, aes(x = bill_length_mm, y = body_mass_g)) +
geom_point() +
geom_smooth(method = "lm") +
labs(
title = "Relation between bill length and body mass",
subtitle = TeX(prep_eq)
)
ggplot(penguins, aes(x = bill_length_mm, y = body_mass_g)) +
geom_point() +
geom_smooth(method = "lm", color = "#ab00fa") +
annotate("text",
label = TeX(prep_eq), x = 35, y = 2500, hjust = 0,
color = "#ab00fa"
) +
labs(title = "Relation between bill length and body mass") +
ylim(2000, 6500)
A more complete overview of {latex2exp} is available through the associated vignette shipped with the package.
Base graphics
In addition to {ggplot2}, we can use {equatiomatic} with base graphics using a similar approach. For example:
plot(penguins$bill_length_mm, penguins$body_mass_g,
pch = 20,
col = "gray80",
xlab = "Bill length",
ylab = "Body Mass",
main = "Relation between bill length and body mass",
sub = TeX(prep_eq)
)
lines(na.omit(penguins$bill_length_mm), predict(m),
col = "#ab00fa"
)
plot(penguins$bill_length_mm, penguins$body_mass_g,
pch = 20,
col = "gray80",
xlab = "Bill length",
ylab = "Body Mass",
main = "Relation between bill length and body mass"
)
lines(na.omit(penguins$bill_length_mm), predict(m),
col = "#ab00fa"
)
text(x = 40, y = 6000, TeX(prep_eq))