equatiomatic

Convert models to LaTeX equations

https://github.com/datalorax/equatiomatic

Keywords from Contributors

Repository

Convert models to LaTeX equations

README.Rmd

---
output: github_document
---



```{r setup, include=FALSE}
knitr::opts_chunk$set(collapse = TRUE, comment = "#>",
  fig.path = "man/figures/README-", out.width = "100%")
```

# equatiomatic 


[![lifecycle](https://lifecycle.r-lib.org/articles/figures/lifecycle-stable.svg)](https://lifecycle.r-lib.org/articles/stages.html#stable) [![R build status](https://github.com/datalorax/equatiomatic/workflows/R-CMD-check/badge.svg)](https://github.com/datalorax/equatiomatic/actions) [![Codecov test coverage](https://codecov.io/gh/datalorax/equatiomatic/branch/master/graph/badge.svg)](https://app.codecov.io/gh/datalorax/equatiomatic?branch=master) [![Covrpage Summary](https://img.shields.io/badge/covrpage-Last_Build_2020_08_24-brightgreen.svg)](https://github.com/datalorax/equatiomatic/tree/master/tests) [![](https://cranlogs.r-pkg.org/badges/grand-total/equatiomatic)](https://cranlogs.r-pkg.org/badges/grand-total/equatiomatic)


The goal of {equatiomatic} is to reduce the pain associated with writing LaTeX code from a fitted model. The package aims to support any model
supported by [{broom}](https://cran.r-project.org/package=broom). See the [introduction to equatiomatic](https://datalorax.github.io/equatiomatic/articles/equatiomatic.html#other-models) for currently supported models.

## Installation

Install from CRAN with:

```{r install-cran, eval=FALSE}
install.packages("equatiomatic")
```

Or get the development version from GitHub with:

```{r install-command, eval=FALSE}
remotes::install_github("datalorax/equatiomatic")
```

## Basic usage

![](https://github.com/datalorax/equatiomatic/raw/master/img/equatiomatic.gif)

The gif above shows the basic functionality.

To convert a model to LaTeX, feed a model object to `extract_eq()`:

```{r example-basic, warning=FALSE, message=FALSE, eval=FALSE}
library(equatiomatic)

# Fit a simple model
mod1 <- lm(mpg ~ cyl + disp, data = mtcars)

# Give the results to extract_eq
extract_eq(mod1)
```

```{r example-basic-preview, echo=FALSE}
library(equatiomatic)
mod1 <- lm(mpg ~ cyl + disp, data = mtcars)

eq1 <- extract_eq(mod1)
print(eq1)
eq1
```

The model can be built in any standard way. It can handle shortcut syntax:

```{r example-shortcut, eval=FALSE}
mod2 <- lm(mpg ~ ., data = mtcars)
extract_eq(mod2)
```

```{r example-shortcut-preview, echo=FALSE}
mod2 <- lm(mpg ~ ., data = mtcars)

eq2 <- extract_eq(mod2)
print(eq2)
eq2
```

When using categorical variables, it will include the levels of the variables as subscripts.

```{r example-categorical, eval=FALSE}
data("penguins", package = "equatiomatic")
mod3 <- lm(body_mass_g ~ bill_length_mm + species, data = penguins)
extract_eq(mod3)
```

```{r example-categorical-preview, echo=FALSE}
mod3 <- lm(body_mass_g ~ bill_length_mm + species, data = penguins)

eq3 <- extract_eq(mod3)
print(eq3)
eq3
```

It helpfully preserves the order the variables are supplied in the formula:

```{r example-preserve-order, eval=FALSE}
set.seed(8675309)
d <- data.frame(cat1  = rep(letters[1:3], 100),
                cat2  = rep(LETTERS[1:3], each = 100),
                cont1 = rnorm(300, 100, 1),
                cont2 = rnorm(300, 50, 5),
                out   = rnorm(300, 10, 0.5))
mod4 <- lm(out ~ cont1 + cat2 + cont2 + cat1, data = d)
extract_eq(mod4)
```

```{r example-preserve-order-preview, echo=FALSE}
set.seed(8675309)
d <- data.frame(cat1  = rep(letters[1:3], 100),
                cat2  = rep(LETTERS[1:3], each = 100),
                cont1 = rnorm(300, 100, 1),
                cont2 = rnorm(300, 50, 5),
                out   = rnorm(300, 10, 0.5))
mod4 <- lm(out ~ cont1 + cat2 + cont2 + cat1, data = d)

eq4 <- extract_eq(mod4)
print(eq4)
eq4
```

## Appearance

You can wrap the equations so that a specified number of terms appear on the right-hand side of the equation using `terms_per_line` (defaults to 4):

```{r example-wrap, eval=FALSE}
extract_eq(mod2, wrap = TRUE)
```

```{r example-wrap-preview, echo=FALSE}
eq2 <- extract_eq(mod2, wrap = TRUE)
print(eq2)
eq2
```

```{r example-wrap-longer, eval=FALSE}
extract_eq(mod2, wrap = TRUE, terms_per_line = 6)
```

```{r example-wrap-longer-preview, echo=FALSE}
eq2 <- extract_eq(mod2, wrap = TRUE, terms_per_line = 6)
print(eq2)
eq2
```

When wrapping, you can change whether the lines end with trailing math operators like `+` (the default), or if they should begin with them using `operator_location = "end"` or `operator_location = "start"`:

```{r example-wrap-longer-location, eval=FALSE}
extract_eq(mod2, wrap = TRUE, terms_per_line = 4, operator_location = "start")
```

```{r example-wrap-longer-location-preview, echo=FALSE}
eq2 <- extract_eq(mod2, wrap = TRUE, terms_per_line = 4,
  operator_location = "start")
print(eq2)
eq2
```

By default, all text in the equation is wrapped in `\operatorname{}`. You can optionally have the variables themselves be italicized (i.e. not be wrapped in `\operatorname{}`) with `ital_vars = TRUE`:

```{r example-italics, eval=FALSE}
extract_eq(mod2, wrap = TRUE, ital_vars = TRUE)
```

```{r example-italics-preview, echo=FALSE}
eq2 <- extract_eq(mod2, wrap = TRUE, ital_vars = TRUE)
print(eq2)
eq2
```


## R Markdown and previewing

If you include `extract_eq()` in an R Markdown chunk, {knitr} will render the equation. If you'd like to see the LaTeX code wrap the call in `print()`.

You can also use the `preview_eq()` function to preview the equation in RStudio:

```{r example-preview, eval=FALSE}
preview_eq(mod1)
```

![](man/figures/preview1.png)

Both `extract_eq()` and `preview_eq()` work with base R or {magrittr} pipes, so you can do something like this:

```{r pipe-example, eval=FALSE}
#library(magrittr)  # if you want to use %>% instead of |>

extract_eq(mod1) |>
  preview_eq()
# Or simply: preview_eq(mod1)
```


## Extra options

There are several extra options you can enable with additional arguments to `extract_eq()`.

### Actual coefficients

You can return actual numeric coefficients instead of Greek letters with `use_coefs = TRUE`:

```{r use-coefs, eval=FALSE}
extract_eq(mod1, use_coefs = TRUE)
```

```{r use-coefs-preview, echo=FALSE}
eq1 <- extract_eq(mod1, use_coefs = TRUE)
print(eq1)
eq1
```

By default, it will remove doubled operators like "+ -", but you can keep those in (which is often useful for teaching) with `fix_signs = FALSE`:

```{r fix-signs, eval=FALSE}
extract_eq(mod1, use_coefs = TRUE, fix_signs = FALSE)
```

```{r fix-signs-preview, echo=FALSE}
eq1 <- extract_eq(mod1, use_coefs = TRUE, fix_signs = FALSE)
print(eq1)
eq1
```

This works in longer wrapped equations:

```{r fix-signs-long, eval=FALSE}
extract_eq(mod2, wrap = TRUE, terms_per_line = 3, use_coefs = TRUE,
  fix_signs = FALSE)
```

```{r fix-signs-long-preview, echo=FALSE}
eq2 <- extract_eq(mod2, wrap = TRUE, terms_per_line = 3, use_coefs = TRUE,
  fix_signs = FALSE)
print(eq2)
eq2
```

## Beyond `lm()`

You're not limited to just `lm` models! {equatiomatic} supports many other models, including logistic regression, probit regression, and ordered logistic regression (with `MASS::polr()`).

### Logistic regression with `glm()`

```{r example-logit, eval=FALSE}
model_logit <- glm(sex ~ bill_length_mm + species, data = penguins,
  family = binomial(link = "logit"))
extract_eq(model_logit, wrap = TRUE, terms_per_line = 3)
```

```{r example-logit-preview, echo=FALSE}
model_logit <- glm(sex ~ bill_length_mm + species, data = penguins,
  family = binomial(link = "logit"))

eq_logit <- extract_eq(model_logit, wrap = TRUE, terms_per_line = 3)
print(eq_logit)
eq_logit
```

### Probit regression with `glm()`

```{r example-probit, eval=FALSE}
model_probit <- glm(sex ~ bill_length_mm + species, data = penguins,
  family = binomial(link = "probit"))
extract_eq(model_probit, wrap = TRUE, terms_per_line = 3)
```

```{r example-probit-preview, echo=FALSE}
model_probit <- glm(sex ~ bill_length_mm + species, data = penguins,
  family = binomial(link = "probit"))

eq_probit <- extract_eq(model_probit, wrap = TRUE, terms_per_line = 3)
print(eq_probit)
eq_probit
```

### Ordered logistic regression with `MASS::polr()`

```{r example-polr, eval=FALSE}
set.seed(1234)
df <- data.frame(
  outcome      = ordered(rep(LETTERS[1:3], 100), levels = LETTERS[1:3]),
  continuous_1 = rnorm(300, 100, 1),
  continuous_2 = rnorm(300, 50, 5))

model_ologit <- MASS::polr(outcome ~ continuous_1 + continuous_2,
  data = df, Hess = TRUE, method = "logistic")
model_oprobit <- MASS::polr(outcome ~ continuous_1 + continuous_2,
  data = df, Hess = TRUE, method = "probit")

extract_eq(model_ologit, wrap = TRUE)
```

```{r example-ologit-preview, echo=FALSE}
set.seed(1234)
df <- data.frame(
  outcome      = ordered(rep(LETTERS[1:3], 100), levels = LETTERS[1:3]),
  continuous_1 = rnorm(300, 100, 1),
  continuous_2 = rnorm(300, 50, 5))

model_ologit <- MASS::polr(outcome ~ continuous_1 + continuous_2,
  data = df, Hess = TRUE, method = "logistic")
model_oprobit <- MASS::polr(outcome ~ continuous_1 + continuous_2,
  data = df, Hess = TRUE, method = "probit")

eq_ologit <- extract_eq(model_ologit, wrap = TRUE)
print(eq_ologit)
eq_ologit
```

```{r example-polr-probit, eval=FALSE}
extract_eq(model_oprobit, wrap = TRUE)
```

```{r example-polr-probit-preview, echo=FALSE}
eq_oprobit <- extract_eq(model_oprobit, wrap = TRUE)
print(eq_oprobit)
eq_oprobit
```


### Ordered regression (logit and probit) with `ordinal::clm()`

```{r example-clm, eval=FALSE}
set.seed(1234)
df <- data.frame(
  outcome      = ordered(rep(LETTERS[1:3], 100), levels = LETTERS[1:3]),
  continuous_1 = rnorm(300, 1, 1),
  continuous_2 = rnorm(300, 5, 5))

model_ologit <- ordinal::clm(outcome ~ continuous_1 + continuous_2,
  data = df, link = "logit")
model_oprobit <- ordinal::clm(outcome ~ continuous_1 + continuous_2,
  data = df, link = "probit")

extract_eq(model_ologit, wrap = TRUE)
```

```{r example-clm-ologit-preview, echo=FALSE}
set.seed(1234)
df <- data.frame(
  outcome      = ordered(rep(LETTERS[1:3], 100), levels = LETTERS[1:3]),
  continuous_1 = rnorm(300, 1, 1),
  continuous_2 = rnorm(300, 5, 5))

model_ologit <- ordinal::clm(outcome ~ continuous_1 + continuous_2,
  data = df, link = "logit")
model_oprobit <- ordinal::clm(outcome ~ continuous_1 + continuous_2,
  data = df, link = "probit")

eq_ologit <- extract_eq(model_ologit, wrap = TRUE)
print(eq_ologit)
eq_ologit
```

```{r example-clm-probit, eval=FALSE}
extract_eq(model_oprobit, wrap = TRUE)
```

```{r example-clm-oprobit-preview, echo=FALSE}
eq_oprobit <- extract_eq(model_oprobit, wrap = TRUE)
print(eq_oprobit)
eq_oprobit
```

## Extension

If you would like to contribute to this package, we'd love your help! We are particularly interested in extending to more models. We hope to support
any model supported by [{broom}](https://cran.r-project.org/package=broom) in the future.

## Code of Conduct

Please note that the 'equatiomatic' project is released with a [Contributor Code of Conduct](https://github.com/datalorax/equatiomatic/blob/master/CODE_OF_CONDUCT.md). By contributing to this project, you agree to abide by its terms.

## A note of appreciation

We'd like to thank the authors of the [{palmerpenguins}](https://allisonhorst.github.io/palmerpenguins/index.html) dataset for generously allowing us to incorporate the `penguins` dataset in our package for example usage.

Horst AM, Hill AP, Gorman KB (2020). *palmerpenguins: Palmer Archipelago (Antarctica) penguin data*. R package version 0.1.0. https://allisonhorst.github.io/palmerpenguins/

Owner

GitHub Events

Committers

Last synced: about 1 year ago

Issues and Pull Requests

Last synced: 10 months ago

Packages