Getting started with R Markdown (Optional)

Learning Objectives

Create a .Rmd document containing R code, text, and plots

Create a YAML header to control output

Understand basic syntax of (R)Markdown

Customise code chunks to control formatting

Use code chunks and in-line code to create dynamic, reproducible documents

R Markdown

R Markdown is a flexible type of document that allows you to seamlessly combine executable R code, and its output, with text in a single document. These documents can be readily converted to multiple static and dynamic output formats, including PDF (.pdf), Word (.docx), and HTML (.html).

The benefit of a well-prepared R Markdown document is full reproducibility. This also means that, if you notice a data transcription error, or you are able to add more data to your analysis, you will be able to recompile the report without making any changes in the actual document.

The rmarkdown package comes pre-installed with RStudio, so no action is necessary.

R Markdown wizard monsters creating a R Markdown document from a recipe. Art by Allison Horst — Image credit: Allison Horst

Creating an R Markdown file

To create a new R Markdown document in RStudio, click File -> New File -> R Markdown:

Screenshot of the New R Markdown file dialogue box in RStudio

Then click on ‘Create Empty Document’. Normally you could enter the title of your document, your name (Author), and select the type of output, but we will be learning how to start from a blank document.

Basic components of R Markdown

To control the output, a YAML (YAML Ain’t Markup Language) header is needed:

---
title: "My Awesome Report"
author: "Emmet Brickowski"
date: ""
output: html_document
---

The header is defined by the three hyphens at the beginning (---) and the three hyphens at the end (---).

In the YAML, the only required field is the output:, which specifies the type of output you want. This can be an html_document, a pdf_document, or a word_document. We will start with an HTML doument and discuss the other options later.

The rest of the fields can be deleted, if you don’t need them. After the header, to begin the body of the document, you start typing after the end of the YAML header (i.e. after the second ---).

Markdown syntax

Markdown is a popular markup language that allows you to add formatting elements to text, such as bold, italics, and code. The formatting will not be immediately visible in a markdown (.md) document, like you would see in a Word document. Rather, you add Markdown syntax to the text, which can then be converted to various other files that can translate the Markdown syntax. Markdown is useful because it is lightweight, flexible, and platform independent.

Some platforms provide a real time preview of the formatting, like RStudio’s visual markdown editor (available from version 1.4).

First, let’s create a heading! A # in front of text indicates to Markdown that this text is a heading. Adding more #s make the heading smaller, i.e. one # is a first level heading, two ##s is a second level heading, etc. upto the 6th level heading.

# Title
## Section
### Sub-section
#### Sub-sub section
##### Sub-sub-sub section
###### Sub-sub-sub-sub section

(only use a level if the one above is also in use)

Since we have already defined our title in the YAML header, we will use a section heading to create an Introduction section.

## Introduction

You can make things bold by surrounding the word with double asterisks, **bold**, or double underscores, __bold__; and italicize using single asterisks, *italics*, or single underscores, _italics_.

You can also combine bold and italics to write something really important with triple-asterisks, ***really***, or underscores, ___really___; and, if you’re feeling bold (pun intended), you can also use a combination of asterisks and underscores, **_really_**, **_really_**.

To create code-type font, surround the word with backticks, `code-type`.

Now that we’ve learned a couple of things, it might be useful to implement them:

## Introduction

This report uses the **tidyverse** package along with the *Portal Project Teaching Database*, 
which has columns that include:

Then we can create a list for the variables using -, +, or * keys.

## Introduction

This report uses the **tidyverse** package along with the *SAFI* dataset, 
which has columns that include:

- record_id
- year
- plot_id
- species_id
- sex
- hindfoot_length
- weight
- genus
- plot_type

You can also create an ordered list using numbers:

1. record_id
2. year
3. plot_id
4. species_id
5. sex
6. hindfoot_length
7. weight
8. genus
9. plot_type

And nested items by tab-indenting:

- record_id
  + Unique identifier of this observation
- year
  + Date of observation
- plot_id
  + Plot where observation was made
- sex
  + The sex of the observed animal

For more Markdown syntax see the following reference guide.

Now we can render the document into HTML by clicking the Knit button in the top of the Source pane (top left), or use the keyboard shortcut Ctrl+Shift+K on Windows and Linux, and Cmd+Shift+K on Mac. If you haven’t saved the document yet, you will be prompted to do so when you Knit for the first time.

The ‘knitting’ process: First, R Markdown is converted to Markdown, which is then converted (via pandoc) to .html, .pdf, .docx, etc.

Writing an R Markdown report

Now we will add some R code from our previous data wrangling and visualisation, which means we need to make sure tidyverse is loaded. It is not enough to load tidyverse from the console, we will need to load it within our R Markdown document. The same applies to our data. To load these, we will need to create a ‘code chunk’ at the top of our document (below the YAML header).

A code chunk can be inserted by clicking Code > Insert Chunk, or by using the keyboard shortcuts Ctrl+Alt+I on Windows and Linux, and Cmd+Option+I on Mac.

The syntax of a code chunk is:

```{r chunk-name}
Here is where you place the R code that you want to run.
```

An R Markdown document knows that this text is not part of the report from the ``` that begins and ends the chunk. It also knows that the code inside of the chunk is R code from the r inside of the curly braces ({}). After the r you can add a name for the code chunk . Naming a chunk is optional, but recommended. Each chunk name must be unique, and only contain alphanumeric characters and -.

To load tidyverse and our surveys_complete.csv file, we will insert a chunk and call it ‘setup’. Since we don’t want this code or the output to show in our knitted HTML document, we add an include = FALSE option after the code chunk name ({r setup, include = FALSE}).

```{r setup, include = FALSE}
library(tidyverse)
library(here)
surveys_complete <- read_csv(here("data/surveys_complete.csv"))
```

If you were unable to complete the previous lesson or did not save the data, then you can create it in a new code chunk instead.

```{r surveys-data, include = FALSE}
surveys_complete <- read_csv(here("data_raw/portal_data_joined.csv")) |>
  drop_na(weight, hindfoot_length, sex) |>
  group_by(species_id) |>
  filter(n() >= 50) |>
  ungroup()
```

Important Note!

The file paths you give in a .Rmd document, e.g. to load a .csv file, are relative to the .Rmd document, not the project root.

We highly recommend the use of the here() function to keep the file paths consistent within your project.

Insert table

Next, we will re-create a table from the Manipulating Data episode which shows average animal weight grouped by species_id and sex. We can do this by creating a new code chunk and calling it ‘weight-tbl’. Or, you can come up with something more creative (just remember to stick to the naming rules).

It isn’t necessary to Knit your document every time you want to see the output. Instead you can run the code chunk with the green triangle in the top right corner of the the chunk, or with the keyboard shortcuts: Ctrl+Alt+C on Windows and Linux, or Cmd+Option+C on Mac.

To make sure the table is formatted nicely in our output document, we will need to use the kable() function from the knitr package. The kable() function takes the output of your R code and knits it into a nice looking HTML table. You can also specify different aspects of the table, e.g. the column names, a caption, etc.

Run the code chunk to make sure you get the desired output.

```{r weight-tbl, message=FALSE}
surveys_complete |>
  group_by(species_id, sex) |>
  summarize(mean_weight = mean(weight, na.rm = TRUE)) |>
  head(n=10) |>
  knitr::kable(caption = "We can also add a caption.",
               col.names = c("Species", "Sex", "Mean Weight"))
```

We can also add a caption.
Species	Sex	Mean Weight
DM	F	41.57354
DM	M	44.32434
DO	F	48.54098
DO	M	49.12102
DS	F	117.39224
DS	M	122.88410
NL	F	154.36964
NL	M	164.87927
OL	F	30.78440
OL	M	31.83369

Customizing chunk output

We mentioned using include = FALSE in a code chunk to prevent the code and output from printing in the knitted document. There are additional options available to customize how the code-chunks are presented in the output document. The options are entered in the code chunk after chunk-nameand separated by commas, e.g. {r chunk-name, eval = FALSE, echo = TRUE}.

Option	Options	Output
`eval`	`TRUE` or `FALSE`	Whether or not the code within the code chunk should be run.
`echo`	`TRUE` or `FALSE`	Choose if you want to show your code chunk in the output document. `echo = TRUE` will show the code chunk.
`include`	`TRUE` or `FALSE`	Choose if the output of a code chunk should be included in the document. `FALSE` means that your code will run, but will not show up in the document.
`warning`	`TRUE` or `FALSE`	Whether or not you want your output document to display potential warning messages produced by your code.
`message`	`TRUE` or `FALSE`	Whether or not you want your output document to display potential messages produced by your code.
`fig.align`	`default`, `left`, `right`, `center`	Where the figure from your R code chunk should be output on the page

Tip

The default settings for the above chunk options are all TRUE.

The default settings can be modified per chunk, or with knitr::opts_chunk$set(),

Entering knitr::opts_chunk$set(echo = FALSE) will change the default of value of echo to FALSE for every code chunk in the document.

Challenge

Play around with the different options in the chunk with the code for the table, and re-Knit to see what each option does to the output.
What happens if you use eval = FALSE and echo = FALSE? What is the difference between this and include = FALSE?

Answer

Create a chunk with {r eval = FALSE, echo = FALSE}, then create another chunk with {r include = FALSE} to compare. eval = FALSE and echo = FALSE will neither run the code in the chunk, nor show the code in the knitted document. The code chunk essentially doesn’t exist in the knitted document as it was never run. Whereas include = FALSE will run the code and store the output for later use.

In-line R code

Now we will use some in-line R code to present some descriptive statistics. To use in-line R-code, we use the same backticks that we used in the Markdown section, with an r to specify that we are generating R-code. The difference between in-line code and a code chunk is the number of backticks. In-line R code uses one backtick (`r`), whereas code chunks use three backticks (``` r ```).

For example, today’s date is `r Sys.Date()`, will be rendered as: today’s date is 2022-09-29.
The code will display today’s date in the output document (well, technically the date the document was last knitted).

The best way to use in-line R code, is to minimise the amount of code you need to produce the in-line output by preparing the output in code chunks. Let’s say we’re interested in presenting the average hindfoot length of a genus.

```{r in-line-code}
# create a summary data frame with the mean hindfoot length by genus
mean_hindfoot <- surveys_complete |>
  group_by(genus) |>
  summarize(mean_hindfoot_length = mean(hindfoot_length))

# and select the genus we want to use
mean_neotoma <- mean_hindfoot %>%
  filter(genus == "Neotoma")
```

Now we can make an informative statement on the means of each village, and include the mean values as in-line R-code. For example:

The average hindfoot length in the genus Neotoma is `r round(mean_neotoma$mean_hindfoot_length, 2)`

becomes …

The average hindfoot length in the genus Neotoma is 32.25.

Because we are using in-line R code instead of the actual values, we have created a dynamic document that will automatically update if we make changes to the dataset and/or code chunks.

Plots

Finally, we will also include a plot, so our document is a little more colorful and a little less boring.

```{r surveys-plot}
yearly_counts <- surveys_complete |>
  count(year, genus)
ggplot(data = yearly_counts, aes(x = year, y = n, color = genus)) +
    geom_line()
```

Challenge

Add a second plot to your document that shows the yearly counts of animals per genera separated by sex and using a facet to create sub plots per genus.
Answer
```{r my-fancy-plot}
 yearly_sex_counts <- surveys_complete |>
   count(year, genus, sex)
 ggplot(data = yearly_sex_counts, mapping = aes(x = year, y = n, color = sex)) +
   geom_line() +
   facet_wrap(facets =  vars(genus))
```

We can also create a caption with the chunk option fig.cap.

```{r chunk-name, fig.cap = "I made this plot while attending an
awesome Data Carpentries workshop where I learned a ton of cool stuff!"}
# Code for plot
```

or, ideally, something more informative.

I made this plot while attending an awesome Data Carpentries workshop where I learned a ton of cool stuff!

Other output options

You can convert R Markdown to a PDF or a Word document (among others). Click the little triangle next to the Knit button to get a drop-down menu. Or you could put pdf_document or word_document in the initial header of the file.

---
title: "My Awesome Report"
author: "Emmet Brickowski"
date: ""
output: word_document
---

Note: Creating PDF documents

Creating .pdf documents may require installation of some extra software. The R package tinytex provides some tools to help make this process easier for R users. With tinytex installed, run tinytex::install_tinytex() to install the required software (you’ll only need to do this once) and then when you Knit to pdf tinytex will automatically detect and install any additional LaTeX packages that are needed to produce the pdf document. Visit the tinytex website for more information.

Resources

Knitr in a knutshell tutorial
Dynamic Documents with R and knitr (book)
R Markdown documentation
R Markdown cheat sheet
Getting started with R Markdown
Markdown tutorial
R Markdown: The Definitive Guide (book by Rstudio team)
Reproducible Reporting
The Ecosystem of R Markdown
Introducing Bookdown

Data Carpentry, 2014-2021.

License. Contributing.

Questions? Feedback? Please file an issue on GitHub.
On Twitter: @datacarpentry

If this lesson is useful to you, consider subscribing to our newsletter or making a donation to support the work of The Carpentries.