Visualisation with ggplot2

# Visualisation with ggplot2
## <a href="https://privefl.github.io/R-presentation/ggplot2.html" class="uri">https://privefl.github.io/R-presentation/ggplot2.html</a>
### Florian Privé
### October 19, 2017 (minor updates in November 2020)

---

class: center, top, inverse
background-image: url(http://hexb.in/vector/ggplot2.svg)
background-position: 50% 80%
background-size: 40%

# http://ggplot2.tidyverse.org/

---

> "The simple graph has brought more information to the data analyst’s mind than any other device." --- John Tukey

---

# Introduction

---

## What does *ggplot2* stand for?

### A __Grammar of Graphics__!

```
ggplot(data = <DATA>) + 
 <GEOM_FUNCTION>(
 mapping = aes(<MAPPINGS>),
 stat = <STAT>, 
 position = <POSITION>
 ) +
 <COORDINATE_FUNCTION> +
 <FACET_FUNCTION>
```

#### You can uniquely describe any plot as a combination of these 7 parameters.

---

## How long have you known **ggplot2**?

<blockquote class="twitter-tweet" data-lang="en" align="center">Happy 10th birthday ggplot2! 🎉🎂📊📈10 years ago today the first version was accepted to CRAN: <a href="https://t.co/tiXIkqnCcA">https://t.co/tiXIkqnCcA</a>&mdash; Hadley Wickham (@hadleywickham) <a href="https://twitter.com/hadleywickham/status/873556949207535616">10 juin 2017</a></blockquote>

---

## Why use ggplot2?

- Automatic legends, colors, etc.

- Easy superposition, facetting, etc.

- Nice rendering (though I don't like the default grey theme).

- Store any ggplot2 object for modification or future recall. Super useful for packages.

- Lots of users (less bugs, more help on Stack Overflow).

- Lots of extensions.

- Nice saving option.

---

# Tidy data?

---

## Untidy data

```r
fertilityData
```

```
## # A tibble: 12 x 7
## Country `1800` `1801` `1802` `1803` `1804` `1805`
## <fct> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 Afghanistan 7 7 7 7 7 7 
## 2 Albania 4.6 4.6 4.6 4.6 4.6 4.6 
## 3 Algeria 6.99 6.99 6.99 6.99 6.99 6.99
## 4 Angola 6.93 6.93 6.93 6.93 6.93 6.93
## 5 Antigua and Barbuda 5 5 4.99 4.99 4.99 4.98
## 6 Argentina 6.8 6.8 6.8 6.8 6.8 6.8 
## 7 Armenia 7.8 7.8 7.81 7.81 7.81 7.82
## 8 Aruba 5.64 5.64 5.64 5.64 5.64 5.64
## 9 Australia 6.5 6.48 6.46 6.44 6.42 6.4 
## 10 Austria 5.1 5.1 5.1 5.1 5.1 5.1 
## 11 Azerbaijan 8.1 8.1 8.1 8.1 8.1 8.1 
## 12 Bahamas 5.9 5.9 5.9 5.9 5.9 5.9
```

The variables are the **country**, **year** and **fertility rate**.

---

## Tidy data

It is easier to work and reason with

- operations

- manipulation

- visualization

Learn more at http://tidyr.tidyverse.org/articles/tidy-data.html.

---

### Tidy the previous data

```r
library(dplyr)
```

```r
(fertilityTidy <- fertilityData %>%
 tidyr::pivot_longer(cols = -Country, names_to = "Year",
 values_to = "Fertility") %>%
 mutate(Year = as.integer(Year)))
```

```
## # A tibble: 72 x 3
## Country Year Fertility
## <fct> <int> <dbl>
## 1 Afghanistan 1800 7 
## 2 Afghanistan 1801 7 
## 3 Afghanistan 1802 7 
## 4 Afghanistan 1803 7 
## 5 Afghanistan 1804 7 
## 6 Afghanistan 1805 7 
## 7 Albania 1800 4.6
## 8 Albania 1801 4.6
## 9 Albania 1802 4.6
## 10 Albania 1803 4.6
## # ... with 62 more rows
```

---

### This is easier to plot

```r
ggplot(data = fertilityTidy) + 
  geom_point(mapping = aes(x = Year, y = Fertility))
```

---

# Basics & Customization

---

### Use black and white theme everywhere

```r
library(ggplot2)
theme_set(theme_bw(18))
```

***

```r
ggplot(data = fertilityTidy) + 
  geom_point(mapping = aes(x = Year, y = Fertility))
```

---

## Drop extra typing

```r
ggplot(fertilityTidy) + 
  geom_point(aes(Year, Fertility))
```

---

## Add colors

```r
ggplot(fertilityTidy) + 
  geom_point(aes(Year, Fertility, color = Country))
```

---

## Add lines: add one geom

```r
ggplot(fertilityTidy) + 
  geom_point(aes(Year, Fertility, color = Country)) + 
  geom_line(aes(Year, Fertility, color = Country))
```

---

### Remove redundancy: move 'aes' to the top

#### So that the mapping is inherited by both geoms

```r
ggplot(fertilityTidy, aes(Year, Fertility, color = Country)) + 
  geom_point() +
  geom_line()
```

---

### Larger points and lines

```r
ggplot(fertilityTidy, aes(Year, Fertility, color = Country)) + 
  geom_point(size = 4) +
  geom_line(size = 3)
```

---

### Futher customization: themes

```r
ggplot(fertilityTidy, aes(Year, Fertility, color = Country)) + 
  geom_point(size = 4) +
  geom_line(size = 3) +
  theme(aspect.ratio = 0.8, legend.key.width = unit(3, "line"))
```

---

# Layers

---

## Iris: base dataset of R

### about plants

```r
as_tibble(iris)   ## print is better
```

```
## # A tibble: 150 x 5
## Sepal.Length Sepal.Width Petal.Length Petal.Width Species
## <dbl> <dbl> <dbl> <dbl> <fct> 
## 1 5.1 3.5 1.4 0.2 setosa 
## 2 4.9 3 1.4 0.2 setosa 
## 3 4.7 3.2 1.3 0.2 setosa 
## 4 4.6 3.1 1.5 0.2 setosa 
## 5 5 3.6 1.4 0.2 setosa 
## 6 5.4 3.9 1.7 0.4 setosa 
## 7 4.6 3.4 1.4 0.3 setosa 
## 8 5 3.4 1.5 0.2 setosa 
## 9 4.4 2.9 1.4 0.2 setosa 
## 10 4.9 3.1 1.5 0.1 setosa 
## # ... with 140 more rows
```

---

## Layers: example with geom_smooth()

```r
ggplot(iris) + 
  geom_point(aes(Petal.Length, Petal.Width, 
                 color = Species, shape = Species), size = 3)
```

---

### Geom_smooth on all: move x and y on top

```r
ggplot(iris, aes(Petal.Length, Petal.Width)) + 
  geom_point(aes(color = Species, shape = Species), size = 3) + 
  geom_smooth(color = "black")
```

```
## `geom_smooth()` using method = 'loess' and formula 'y ~ x'
```

---

### Points on top: change the order of layers

```r
ggplot(iris, aes(Petal.Length, Petal.Width)) + 
  geom_smooth(color = "black") + 
  geom_point(aes(color = Species, shape = Species), size = 3)
```

```
## `geom_smooth()` using method = 'loess' and formula 'y ~ x'
```

---

### Geom_smooth by group

```r
ggplot(iris, aes(Petal.Length, Petal.Width)) + 
  geom_smooth(aes(group = Species), color = "black") + 
  geom_point(aes(color = Species, shape = Species), size = 3)
```

```
## `geom_smooth()` using method = 'loess' and formula 'y ~ x'
```

---

### Or use color for both geoms

```r
ggplot(iris, aes(Petal.Length, Petal.Width, color = Species)) + 
  geom_smooth() + 
  geom_point(aes(shape = Species), size = 3)
```

```
## `geom_smooth()` using method = 'loess' and formula 'y ~ x'
```

---

## An important application

### Simpson's Paradox

---

## Learn more

- [Chapter *Data Visualisation*](http://r4ds.had.co.nz/data-visualisation.html) of

---

## Find answers

<blockquote class="twitter-tweet" data-lang="en" align="center">ok, hands up... when looking up code errors/questions with the intent of finding answers on Stack Overflow, do you:&mdash; Sharla Gelfand (@sharlagelfand) <a href="https://twitter.com/sharlagelfand/status/915654253456318464?ref_src=twsrc%5Etfw">4 octobre 2017</a></blockquote>
<script async src="//platform.twitter.com/widgets.js" charset="utf-8"></script>

<center>e.g. google "larger points legend ggplot2"

---

## Using the Stack Overflow answer

```r
ggplot(iris) + 
  geom_point(aes(Petal.Length, Petal.Width, 
                 color = Species, shape = Species), size = 3) +
  guides(colour = guide_legend(override.aes = list(size = 10)))
```

---

## Go check the [**R Graph Gallery**](https://www.r-graph-gallery.com/ggplot2-package.html)

<blockquote class="twitter-tweet" data-cards="hidden" data-lang="en" align="center">🍾🍾 Today the <a href="https://twitter.com/hashtag/rstats?src=hash&amp;ref_src=twsrc%5Etfw">#rstats</a> graph gallery reached 1.000.000 visits! 🍾🍾 Thanks to all users &amp; contributors!<a href="https://t.co/94JzuHDJot">https://t.co/94JzuHDJot</a> <a href="https://t.co/xu4XIsqmut">pic.twitter.com/xu4XIsqmut</a>&mdash; The R Graph Gallery (@R_Graph_Gallery) <a href="https://twitter.com/R_Graph_Gallery/status/915092693977460741?ref_src=twsrc%5Etfw">3 octobre 2017</a></blockquote>
<script async src="//platform.twitter.com/widgets.js" charset="utf-8"></script>

---

# More

---

### Coordinates

```r
ggplot(iris) + 
  geom_point(aes(Petal.Length, Petal.Width, 
                 color = Species, shape = Species), size = 3) + 
  scale_x_log10(breaks = 1:7)
```

---

### Facets

```r
ggplot(iris) + 
  geom_point(aes(Petal.Length, Petal.Width, 
                 color = Species, shape = Species), size = 3) + 
  facet_grid(~ Species)
```

---

## Iterate over variables with **aes_string**

```r
(var <- names(iris)[1:4])
```

```
## [1] "Sepal.Length" "Sepal.Width"  "Petal.Length" "Petal.Width"
```

```r
p_list <- list()
for (i in seq_along(var)) {
 p_list[[i]] <- ggplot(iris, coeff = 0.6) + 
 geom_density(aes_string(var[i], fill = "Species"), alpha = 0.6)
 
}

str(p_list, max.level = 1)
```

```
## List of 4
##  $ :List of 9
##   ..- attr(*, "class")= chr [1:2] "gg" "ggplot"
##  $ :List of 9
##   ..- attr(*, "class")= chr [1:2] "gg" "ggplot"
##  $ :List of 9
##   ..- attr(*, "class")= chr [1:2] "gg" "ggplot"
##  $ :List of 9
##   ..- attr(*, "class")= chr [1:2] "gg" "ggplot"
```

---

## Combine plots with [cowplot](https://cran.r-project.org/web/packages/cowplot/vignettes/introduction.html)

```r
cowplot::plot_grid(plotlist = p_list, ncol = 2, align = "hv",
                   labels = LETTERS[1:4], label_size = 15)
```

---

### Common legend

```r
lapply(p_list, function(p) p + theme(legend.position = "none")) %>%
  cowplot::plot_grid(plotlist = ., ncol = 2, align = "hv",
                   labels = LETTERS[1:4], label_size = 15) %>%
  cowplot::plot_grid(cowplot::get_legend(p_list[[1]]),
                     rel_widths = c(1, 0.3))
```

---

### Your turn

Create a similar plot by pivoting the data + using facets

```r
(iris_tidy <- tidyr::pivot_longer(iris, -Species))
```

```
## # A tibble: 600 x 3
## Species name value
## <fct> <chr> <dbl>
## 1 setosa Sepal.Length 5.1
## 2 setosa Sepal.Width 3.5
## 3 setosa Petal.Length 1.4
## 4 setosa Petal.Width 0.2
## 5 setosa Sepal.Length 4.9
## 6 setosa Sepal.Width 3 
## 7 setosa Petal.Length 1.4
## 8 setosa Petal.Width 0.2
## 9 setosa Sepal.Length 4.7
## 10 setosa Sepal.Width 3.2
## # ... with 590 more rows
```

---

```r
ggplot(iris_tidy) +
  geom_density(aes(value, fill = Species), alpha = 0.6) +
  facet_wrap(~ name, scales = "free")
```

---

## Interactive plots

```r
ggplot(iris, aes(Petal.Length, Petal.Width, 
                   color = Species, shape = Species)) + 
  geom_point(size = 3)
```

---

## Transform ggplot to plotly

```r
plotly::ggplotly(width = 700, height = 450)
```

### Add more infos

```r
plotly::ggplotly(
  last_plot() + aes(text = bigstatsr::asPlotlyText(iris)),
  tooltip = "text", width = 700, height = 420)
```

You might want to look at package [{widgetframe}](https://cran.r-project.org/web/packages/widgetframe/vignettes/Using_widgetframe.html).

---

## Miscellaneous

- [Pie charts](https://guangchuangyu.github.io/2016/12/scatterpie-for-plotting-pies-on-ggplot/) but [others plots are often better](http://annkemery.com/pie-chart-guidelines/)

- [Spatial Visualization](https://cran.r-project.org/web/packages/ggmap/index.html)

- [Heatmaps](http://blog.aicry.com/r-heat-maps-with-ggplot2/)

- [Cookbook for R - Graphs](http://www.cookbook-r.com/Graphs/)

- [**Cheatsheet**](https://raw.githubusercontent.com/rstudio/cheatsheets/main/data-visualization.pdf)

- [Top 50 ggplot2 Visualizations](http://r-statistics.co/Top50-Ggplot2-Visualizations-MasterList-R-Code.html)

- [**Viridis color palette**](https://cran.r-project.org/web/packages/viridis/vignettes/intro-to-viridis.html)

- [An RStudio addin for ggplot2 theme tweaking](https://github.com/calligross/ggthemeassist)

- [**Publication Ready Plots**](http://www.sthda.com/english/rpkgs/ggpubr/)

- [Extensions](https://exts.ggplot2.tidyverse.org/)

---

# Two last points from my experience

---

## Why I finally switched to using ggplot2?

```r
library(bigstatsr)
X <- big_attachExtdata()
svd <- big_SVD(X, big_scale(), k = 10)
plot(svd, type = "scores")
```

---

### An object that the user can modify

```r
pop <- rep(c("POP1", "POP2", "POP3"), c(143, 167, 207))
last_plot() + 
 # add colors
 aes(color = pop) + labs(color = "Population") +
 ## change the place of the legend
 theme(legend.position = c(0.85, 0.2)) + 
 ## change the title and the label of the x-axis
 labs(title = "Yet another title", x = "with a new 'x' label")
```

---

## How I choose the size of a plot?

- I plot something, e.g.

```r
ggplot(iris, aes(Petal.Length, Petal.Width, 
                 color = Species, shape = Species)) + 
  geom_point(size = 3)
```

- I use the *Zoom* button of RStudio

- I resize the "Plot Zoom" windows till I'm satisfied

- Right-click on this window -> "Open image" (or "Inspect element")

- Then, in `ggsave()`, I use the dimensions that are displayed with `scale = 1/100` (calibrate the value for your computer), e.g.

```r
ggsave("myggplot.pdf", scale = 1/100, width = 888, height = 725)
```

---

# Your turn

---

## Use this data..

```r
(df <- dplyr::filter(gapminder::gapminder, year == 1992))
```

```
## # A tibble: 142 x 6
## country continent year lifeExp pop gdpPercap
## <fct> <fct> <int> <dbl> <int> <dbl>
## 1 Afghanistan Asia 1992 41.7 16317921 649.
## 2 Albania Europe 1992 71.6 3326498 2497.
## 3 Algeria Africa 1992 67.7 26298373 5023.
## 4 Angola Africa 1992 40.6 8735988 2628.
## 5 Argentina Americas 1992 71.9 33958947 9308.
## 6 Australia Oceania 1992 77.6 17481977 23425.
## 7 Austria Europe 1992 76.0 7914969 27042.
## 8 Bahrain Asia 1992 72.6 529491 19036.
## 9 Bangladesh Asia 1992 56.0 113704579 838.
## 10 Belgium Europe 1992 76.5 10045622 25576.
## # ... with 132 more rows
```

---

## ..to reproduce this plot

---

```r
ggplot(df) + 
  geom_point(aes(gdpPercap, lifeExp, size = pop / 1e6, 
                 color = continent)) +
  scale_x_log10(breaks = c(300, 1e3, 3e3, 10e3, 30e3)) + 
  labs(title = "Gapminder for 1992",
       x = "Gross Domestic Product (log scale)",
       y = "Life Expectancy at birth (years)",
       color = "Continent", size = "Population\n(millions)")
```

---

# Thanks!

Presentation available at

https://privefl.github.io/R-presentation/ggplot2.html

Twitter and GitHub: [@privefl](https://twitter.com/privefl)