Context

Writing R Markdown document makes possible to insert R code and its results in a report with a choosen output format (HTML, PDF, Word). It is a real asset for analysis reproducibility as well as communication of methods and results. Doing daily data analysis, I usually deliver outputs in report and R Markdown naturally became an essential tool of my workflow.

Data analysis without data visualisation is like playing darts in the dark, there is a good chance you’ll miss the bullseye point. You’ll find quite a few R packages to build graphics but I have a preference for ggplot2 (I’m not alone!). Therefore, ggplot2 graphics are often included in my R Markdown documents.

Features of both packages are highly flexible and you CAN always get what you want ! But if you are just starting out, getting what you want can be cumbersome. In this post, I share with you some tips found over time.

Requirements

1. I assume you have already made a graphic with ggplot2 or at least seen some ggplot2 code. If not, you can have a look at this book freely available online. To avoid iris data, I will use a data visualisation of Palmer penguins data recently included in a R package by Allison Horst (go see her illustrations too !). I start with this graphic :

library(ggplot2)
p <- ggplot(penguins_raw, aes(x = culmen_length_mm, y = culmen_depth_mm, color = species, fill = species)) +
  geom_point() +
  geom_smooth(method = "lm", formula = "y ~ x", alpha = 0.3) +
  scale_color_viridis_d() +
  scale_fill_viridis_d() +
  labs(x = "Culmen Length (mm)", y = "Culmen depth (mm)", fill = "Species", color = "Species")
p

2. Besides, it’s better if you know how to create a R Markdown document and you know how to include R code in it (with a chunk). If not, start here.

Making your own ggplot2 theme

ggplot2 theme manages how your graphic looks like. All elemements can be changed through the theme() function but there also are pre-configured. The default theme used by ggplot2 is theme_gray() but I often switch for theme_bw() (for black and white).

p + labs(title = "Thème par défaut")
p + theme_bw() + labs(title = "Avec theme_bw()")

If you always use the same modifications with theme() function, I highly suggest that you create your own theme. Here is a customised one :

theme_ben <- function(base_size = 14) {
  theme_bw(base_size = base_size) %+replace%
    theme(
      # L'ensemble de la figure
      plot.title = element_text(size = rel(1), face = "bold", margin = margin(0,0,5,0), hjust = 0),
      # Zone où se situe le graphique
      panel.grid.minor = element_blank(),
      panel.border = element_blank(),
      # Les axes
      axis.title = element_text(size = rel(0.85), face = "bold"),
      axis.text = element_text(size = rel(0.70), face = "bold"),
      axis.line = element_line(color = "black", arrow = arrow(length = unit(0.3, "lines"), type = "closed")),
      # La légende
      legend.title = element_text(size = rel(0.85), face = "bold"),
      legend.text = element_text(size = rel(0.70), face = "bold"),
      legend.key = element_rect(fill = "transparent", colour = NA),
      legend.key.size = unit(1.5, "lines"),
      legend.background = element_rect(fill = "transparent", colour = NA),
      # Les étiquettes dans le cas d'un facetting
      strip.background = element_rect(fill = "#17252D", color = "#17252D"),
      strip.text = element_text(size = rel(0.85), face = "bold", color = "white", margin = margin(5,0,5,0))
    )
}

And the results compared to the default one :

p + labs(title = "Thème par défaut")
p + theme_ben() + labs(title = "Avec mon thème personnalisé")

Building a customised theme is done by creating a R function where a pre-configured theme is used but some elements are modified with the theme() function. The tip is in the use of %+replace% in place of the classic +. unlike the latter, %+replace% doesn’t only update elements of a theme but replaces them entirely.

Now you just have to let your creativity flows. There are lots of editable elements so the customisation is pretty much limitless. You can find the list of elements in the webpage of the theme() function. In the example, the modified elements are for the whole figure (plot.* elements), the grapihic area (pnael.*), the axis (axis.*), the legend (legend.*) and the graphics labels (strip.*) when facetting is used.

In my opinion, axis and legends are essential elements so my choices go towards highlighted them through their relative size using rel() function which return a proportion a the base size (base_size) and bolding theme (face = "bold").

Customisation of a ggplot2 theme can first be hard work but if you are going to often use the same configurations, it’s worth it. There are plenty ressources on the web.You can also contact me, I’ll be glad to help.

You can now insert your theme in a chunk at the beginning of your R Markdown document to use it all along. You can also create a R package with your theme, among others, and load this package.

Your ggplot figure in R Markdown

```{r option1 = valeur_option1, option2 = valeur_option2}
# Le code R ici
```

```{r fig.asp = 0.8, fig.width = 3}
p + theme_ben()
```

```{r fig.asp = 0.8, fig.width = 10}
p + theme_ben()
```

```{r fig.asp = 0.8, fig.width = 7}
p + theme_ben()
```

```{r fig.asp = 0.8, fig.width = 9}
p + theme_ben(base_size = 14)
```

```{r fig.asp = 0.8, fig.width = 7}
p + theme_ben(base_size = 12)
```

```{r fig.asp = 0.8, fig.width = 7, out.width = "100%}
p + theme_ben()
```

```{r fig.asp = 0.8, fig.width = 7, out.width = "50%}
p + theme_ben()
```

Don’t repeat yourself

it’s kind of annoying to write several time the exact same thing. Here, for every chunk with a ggplot2 figure, you need to tell that you want it with your newly customised theme and you have to configure chunk options each time. Don’t worry, solutions to deal with your (my!) lazyness exist.

theme_set(theme_ben()) # change theme_ben() with the wanted theme

```{r setup, include=FALSE}
knitr::opts_chunk$set(
 fig.width = 6,
 fig.asp = 0.8,
 out.width = "80%"
)
```

---
title: "My awesome title"
author: "Benjamin Louis"
date: "25/06/2020"
output: html_document
---


```{r setup, include=FALSE}
# Chunk options
knitr::opts_chunk$set(
 fig.width = 6,
 fig.asp = 0.8,
 out.width = "80%"
)
```

```{r theme_ggplot2, echo = FALSE}
# Creating a ggplot2 theme
library(ggplot)
theme_ben <- function(base_size = 14) {
  theme_bw(base_size = base_size) %+replace%
    theme(
      # changed theme options
    )
}
# Changing the default theme
theme_set(theme_ben())
```