ggplot2_Grammar_Adv-Layers.Rmd

---
title: "ggplot2 & Grammar of Graphics<br><br>The Four Advanced Layers"
subtitle: "Facets, Statistical Transformations,<br>Coordinate Systems, Themes"
author: "StatistikinDD"
date: "Created: `r Sys.Date()`"
output:
  xaringan::moon_reader:
    chakra: libs/remark-latest.min.js
    lib_dir: libs
    css: ["libs/_css/xaringan-themer.css", "libs/_css/my_css.css"]
    nature:
      highlightStyle: github
      highlightLines: true
      countIncrementalSlides: false
      slideNumberFormat: "%current%"
      ratio: 16:9
---

```{r setup, include = FALSE}

options(htmltools.dir.version = FALSE)
knitr::opts_chunk$set(echo = FALSE, warning = FALSE, message = FALSE, comment = "")

library(tidyverse)
library(ggthemes)

# From https://cran.r-project.org/web/packages/xaringanthemer/vignettes/xaringanthemer.html
library(xaringanthemer)
style_mono_accent(
  base_color = "#1c5253",
  header_font_google = google_font("Josefin Sans"),
  text_font_google   = google_font("Montserrat", "300", "300i"),
  code_font_google   = google_font("Fira Mono"),
  outfile = "libs/_css/xaringan-themer.css"
)

theme_set(theme_gray(base_size = 18))

```

# ggplot2: Layers

.pull-left[
Next to the necessary basic layers, ...

.content-box-grey[
**1. Data**  
**2. Aesthetics**  
**3. Geometries**  
 ]

... there are four advanced / optional layers:

.content-box-gray[
**4. Facets: Display subgroups in separate plotting areas**  
**5. Statistical transformations**  
**6. Coordinate systems**  
**7. Themes: *Non data ink* **  
 ]
]

---

# Our Starting Point

## A Basic Scatterplot

.pull-left[
```{r more-aes, echo = TRUE, eval = FALSE}

ggplot(diamonds, aes(x = carat, y = price)) + #<<
  geom_point(size = 0.5, alpha = 0.5) + #<<
  labs(title = "Diamonds: Price by Carat (weight)",
       x = "Carat", y = "Price",
       caption = "Diamonds dataset, ggplot2")

```

* To display over 50.000 data points, we use both a reduced point size and some opacity

* Note that both price and carat are heavily right-skewed

]

.pull-right[
```{r more-aes-exec, echo = FALSE, eval = TRUE, ref.label = "more-aes", cache = TRUE}
```
]

---

# Facets (Layer 4)

.pull-left[
```{r facets, echo = TRUE, eval = FALSE}

ggplot(diamonds, aes(x = carat, y = price)) +
  geom_point(size = 0.5, alpha = 0.5) +
  facet_wrap(vars(cut)) + #<<
  labs(title = "Diamonds: Price by carat (weight)",
       x = "Carat", y = "Price",
       caption = "Diamonds dataset, ggplot2")

```

* Specify layout using the  
**nrow** and **ncol** parameters
* You can use the formula interface:  
**facet_wrap(~ cut)**
* You can facet by two variables
* Alternative to **facet_wrap()**: **facet_grid()**

]

.pull-right[
```{r facets-exec, echo = FALSE, eval = TRUE, ref.label = "facets", cache = TRUE}
```
]

---

# Statistical transformations (Layer 5)

## Adding a Linear Smooth (lm)

.pull-left[
```{r stats-lm, echo = TRUE, eval = FALSE}

ggplot(diamonds, aes(x = carat, y = price)) +
  geom_point(size = 0.5, alpha = 0.5) +
  geom_smooth(method = "lm") + #<<
  labs(title = "Diamonds: Price by Carat (weight)",
       subtitle = "Smoothing Method: lm", #<<
       x = "Carat", y = "Price",
       caption = "Diamonds dataset, ggplot2")

```

* Check out the **stat_xxx()** functions
* Use case: *stat_summary()* to show means in boxplots

]

.pull-right[
```{r stats-lm-exec, echo = FALSE, eval = TRUE, ref.label = "stats-lm", cache = TRUE}
```
]

---

# Statistical transformations (Layer 5)

## Adding a Nonlinear Smooth (gam)

```{r, echo = FALSE}
theme_set(theme_gray(base_size = 18))
```


.pull-left[
```{r stats-gam, echo = TRUE, eval = FALSE}

ggplot(diamonds, aes(x = carat, y = price)) +
  geom_point(size = 0.5, alpha = 0.5) +
  geom_smooth(method = "gam") + #<<
  labs(title = "Diamonds: Price by Carat (weight)",
       subtitle = "Smoothing Method: gam", #<<
       x = "Carat", y = "Price",
       caption = "Diamonds dataset, ggplot2")

```

GAM = Generalized Additive Model

]

.pull-right[
```{r stats-gam-exec, echo = FALSE, eval = TRUE, ref.label = "stats-gam", cache = TRUE}
```
]

---

# Coordinate Systems (Layer 6)

## Use logarithmic axes, label price in dollars

.pull-left[
```{r stats-coord, echo = TRUE, eval = FALSE}

ggplot(diamonds, aes(x = carat, y = price)) +
  geom_point(size = 0.5, alpha = 0.5) +
  geom_smooth(method = "gam") +
  scale_x_log10() + #<<
  scale_y_log10(labels = scales::dollar) + #<<
  labs(title = "Diamonds: Price by Carat (weight)",
       subtitle = "Axes on Base 10 Log Scale\nSmoothing Method: gam",
       x = "Carat", y = "Price",
       caption = "Diamonds dataset, ggplot2")

```

* More options for transformations:  
*scale_x_continuous(trans = ...)*
* Create custom transformations via *scales::trans_new()*

]

.pull-right[
```{r stats-coord-exec, echo = FALSE, eval = TRUE, ref.label = "stats-coord", cache = TRUE}
```
]

---

# Themes (Layer 7)

## Theme Classic

```{r ggplot-p, echo = FALSE, eval = TRUE}

p <- ggplot(diamonds, aes(x = carat, y = price)) +
  geom_point(size = 0.5, alpha = 0.5) +
  geom_smooth(method = "gam") +
  scale_x_log10() + #<<
  scale_y_log10(labels = scales::dollar) + #<<
  labs(title = "Diamonds: Price by Carat (weight)",
       subtitle = "Axes on Base 10 Log Scale\nSmoothing Method: gam",
       x = "Carat", y = "Price",
       caption = "Diamonds dataset, ggplot2")

# Increasing font sizes

```

.pull-left[
```{r theme-classic, echo = TRUE, eval = FALSE, cache = TRUE}

p + theme_classic()

```

* We define the plot using `p <- ggplot(...)`.
* Then we can simply define a theme using `p + theme_xxx()`.

]

.pull-right[
```{r theme-classic-exec, echo = FALSE, eval = TRUE, cache = TRUE}

p + theme_classic(base_size = 18)

```

]

---

# Themes (Layer 7)

## Theme Dark

```{r, echo = FALSE}
theme_set(theme_dark(base_size = 18))
```

.pull-left[
```{r theme-dark, echo = TRUE, eval = FALSE, cache = TRUE}

p + theme_dark()

```

See more themes: Type *?theme_* in RStudio

]

.pull-right[
```{r theme-dark-exec, echo = FALSE, eval = TRUE, cache = TRUE}
p + theme_dark(base_size = 18)
```
]

---

# Themes (Layer 7)

## More pre-defined themes: ggthemes

.pull-left[
```{r theme-wsj, echo = TRUE, eval = FALSE, cache = TRUE}

library(ggthemes)

p + theme_wsj()

```

* theme_wsj(): Based on plots in  
*The Wall Street Journal*

* More themes: See packages  
**hrbrthemes** and **ggtech**

]

.pull-right[
```{r theme-wsj-exec, echo = FALSE, eval = TRUE, cache = TRUE}
p + theme_wsj()
```

]

---

# Themes (Layer 7)

## A Simple Theme Customization

```{r, echo = FALSE}
theme_set(theme_gray(base_size = 18))
```

.pull-left[
```{r theme-custom, echo = TRUE, eval = FALSE}

p + theme(
  panel.background = element_blank(),
  panel.grid = element_blank(),
  axis.text = element_text(colour = "darkblue"))

```

* Starting point: Default theme theme_gray()
* Remove gray background
* Remove grid lines
* Change font color for axes labels to dark blue

]

.pull-right[
```{r theme-custom-exec, echo = FALSE, eval = TRUE, ref.label = "theme-custom", cache = TRUE}

```
]

---

class: center, middle

# Thanks!

### Youtube: StatistikinDD

### Twitter: @StatistikinDD

### github: fjodor

Slides created via the R package [**xaringan**](https://github.com/yihui/xaringan).

The chakra comes from [remark.js](https://remarkjs.com), [**knitr**](https://yihui.org/knitr), and [R Markdown](https://rmarkdown.rstudio.com).

Thanks to **Yihui Xie** for *{knitr}* and *{xaringan}* and **Garrick Aden-Buie** for *{xaringanthemer}*.