ggplot2 Concepts

class: center, middle, inverse, title-slide

.title[
# ggplot2 Concepts
]

---

class: inverse
## Outline

### 1. A layered grammar

### 2. Geometrical layers
  
### 3. Statistical layers

### 4. Facets

### 5. Storing a ggplot as an object

---
## A layered grammar of graphics

- ggplot2 works on the philosophy of adding layers to the visualization

- there are 7 components of a plot

.center[
<img src="images/layers2.png" width="500">
]

---

## A layered grammar of graphics

- only 3 of these components are necessary to make a layer:

- **data** are the subjects & objects of the data visualization  
  
  - **Aesthetic mappings** (aes) substitute visual properties (aesthetics) for the data

- The **geom** is what relates the data to a visual element.

- these three components allow us maximum flexibility to make subtle changes in each layer to clearly communicate our message.

.center[
 
<img src="images/layers1.png" width="500">
]

---
## The grammar of `{ggplot2}`

---

## The grammar of `{ggplot2}`

<table style='width:100%;font-size:14pt;'>
 <tr>
 <th>Component</th>
 <th>Function</th>
 <th>Explanation</th>
 </tr>
 <tr>
 <td>Data</td>
 <td><code>ggplot(data)</code>&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</td>
 <td style='font-style:italic;'>The raw data that you want to visualise.</td>
 </tr>
 <tr>
 <td>Aesthetics&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;</td>
 <td><code>aes()</code></td>
 <td style='font-style:italic;'>Aesthetic mappings between variables and visual properties.</td>
 <tr>
 <td>Geometries</td>
 <td><code>geom_*()</code></td>
 <td style='font-style:italic;'>The geometric shapes representing the data.</td>
 </tr>
 <tr>
 <td>Statistics</td>
 <td><code>stat_*()</code></td>
 <td style='font-style:italic;'>The statistical transformations applied to the data.</td>
 </tr>
 <tr>
 <td>Scales</td>
 <td><code>scale_*()</code></td>
 <td style='font-style:italic;'>Maps between the data and the aesthetic dimensions.</td>
 </tr>
 <tr>
 <td>Coordinate System</td>
 <td><code>coord_*()</code></td>
 <td style='font-style:italic;'>The positioning of the data in a 2D data visualization.</td>
 </tr>
 <tr>
 <td>Facets</td>
 <td><code>facet_*()</code></td>
 <td style='font-style:italic;'>The arrangement of the data into a grid of plots.</td>
 </tr>
 <tr>
 <td>Themes</td>
 <td><code>theme()</code> and <code>theme_*()</code></td>
 <td style='font-style:italic;'>The overall visual defaults of a plot.</td>
 </tr>
</table>

---
class: inverse, center, middle

# Example

---
## The Data

### OECD's Program for International Student Assessment (PISA)

- aims to inform educational policies and practices
- 540,000 15-year-olds from 72 participating countries and economies
- designed to gauge mastery of key subjects (math, science, reading)
- background questionnaire with questions about themselves, their family and home, and their school and learning experiences

```r
## devtools::install_github("haleyjeppson/NCME23data")
library(NCME23data)
## weighted US sample of size 99
data(pisa_usa)
## weighted sample of 1000
data(pisa_small)
```

---
## Data Variables

```
##  [1] "country"                  "OECD"                    
##  [3] "id"                       "weight"                  
##  [5] "sex"                      "grade"                   
##  [7] "computer"                 "software"                
##  [9] "internet"                 "addit_time_math"         
## [11] "addit_time_science"       "parent_support"          
## [13] "parent_status"            "want_best_grades"        
## [15] "want_best_student"        "test_anxiety"            
## [17] "enjoy_cooperation"        "sense_of_belonging"      
## [19] "parent_support_emotional" "HOMESCH"                 
## [21] "ENTUSE"                   "ICTHOME"                 
## [23] "ICTSCH"                   "wealth"                  
## [25] "parent_edu"               "learning_mins"           
## [27] "escs_index"               "teacher_support_science" 
## [29] "teacher_direct_science"   "inquiry_based_science"   
## [31] "science_self_efficacy"    "math"                    
## [33] "reading"                  "science"                 
## [35] "learning_hours"           "region"
```

---
## Data

.left-code[

```r
*ggplot(data = pisa_usa)
```
]
.right-plot[
<img src="ggplot-concepts_files/figure-html/unnamed-chunk-3-1.png" width="100%" />
]

---
## Aesthetic Mapping

### Link variables in data to graphical properties

- **Axes**: `x`, `y`

- **Grouping**: `group`

- **Other visual properties**: `color`, `fill`, `alpha` (transparency), `size`, `shape`, `linetype`

- **Other**: `weight`, `z`, `xmin`, `xmax`, `ymin`, `ymax`, ...

---
## Aesthetic Mapping

Use the `aes()` function inside `ggplot()`

.left-code[

```r
ggplot(data = pisa_usa,
* mapping = aes(x = math, y = reading))
```
]
.right-plot[
<img src="ggplot-concepts_files/figure-html/unnamed-chunk-4-1.png" width="100%" />
]

---
## Aesthetic Mapping

Use implicit matching

.left-code[

```r
ggplot(pisa_usa,
* aes(x = math, y = reading))
```
]
.right-plot[
<img src="ggplot-concepts_files/figure-html/unnamed-chunk-5-1.png" width="100%" />
]

---
class: inverse, center, middle

# Geometrical Layers

---
## Geometries

### How to interpret aesthetics as graphical representations

![](ggplot-concepts_files/figure-html/unnamed-chunk-6-1.png)

???

control the way that data is displayed

---
## Geometries

We build up a data visualization in ggplot2 with the `+` operator.

.left-code[

```r
ggplot(
 pisa_usa, 
 aes(x = math, y = reading)
* ) +
* geom_point()
```
]
.right-plot[
<img src="ggplot-concepts_files/figure-html/unnamed-chunk-7-1.png" width="100%" />
]
---
## Geometries

Define mappings for a particular geom only

.left-code[

```r
ggplot(pisa_usa) +
* geom_point(aes(x = math, y = reading))
```
]
.right-plot[
<img src="ggplot-concepts_files/figure-html/unnamed-chunk-8-1.png" width="100%" />
]

---
## Geometries

Define data for a particular geom only

.left-code[

```r
*ggplot() +
* geom_point(data = pisa_usa,
 aes(x = math, y = reading))
```
]
.right-plot[
<img src="ggplot-concepts_files/figure-html/unnamed-chunk-9-1.png" width="100%" />
]

---
## Visual Properties of Layers

The `geom_*()` suite of functions can take many arguments, which vary by the geom type

.left-code[

```r
ggplot(
 pisa_usa, 
 aes(x = math, y = reading)
 ) +
 geom_point(
* color = "#3C5488",
* alpha = .7,
* shape = 17,
* stroke = 1,
* size = 5
 )
```
]
.right-plot[
<img src="ggplot-concepts_files/figure-html/unnamed-chunk-10-1.png" width="100%" />
]

---
## Setting vs Mapping Visual Properties

.tall[
.pull-left[

```r
ggplot(
    pisa_usa, 
    aes(x = math, y = reading)
  ) +
  geom_point(
*   color = "#3C5488",
    alpha = .7
  )
```

![](ggplot-concepts_files/figure-html/unnamed-chunk-11-1.png)
]
.pull-right[

```r
ggplot(
    pisa_usa, 
    aes(x = math, y = reading)
  ) +
  geom_point(
*   aes(color = sex),
    alpha = .7
  )
```

![](ggplot-concepts_files/figure-html/unnamed-chunk-12-1.png)
]

]

---
## Local vs. Global Encoding

.tall[
.pull-left[

```r
ggplot(
    pisa_usa, 
*   aes(x = math, y = reading)
  ) +
  geom_point(
*   aes(color = sex),
    alpha = .7
  )
```

![](ggplot-concepts_files/figure-html/unnamed-chunk-13-1.png)
]
.pull-right[

```r
ggplot(
    pisa_usa, 
*   aes(x = math, y = reading,
*       color = sex),
  ) +
  geom_point(
    alpha = .7
  )
```

![](ggplot-concepts_files/figure-html/unnamed-chunk-14-1.png)
]
]

---
## Other geoms

There are many types of geoms and their mapping requirements differ

.tall[
.pull-left[

```r
ggplot(pisa_usa) +
* geom_point(
*   aes(x = math, y = reading),
    color = "#3C5488", 
    alpha = .7)
```

![](ggplot-concepts_files/figure-html/unnamed-chunk-15-1.png)
]
.pull-right[

```r
ggplot(pisa_usa) +
* geom_density(
*   aes(x = math),
    color = "#3C5488", 
    fill = "#3C5488",
    alpha = .7)
```

![](ggplot-concepts_files/figure-html/unnamed-chunk-16-1.png)
]
]

---
## Other geoms

There are many types of geoms and their mapping requirements differ

.tall[
.pull-left[

```r
ggplot(pisa_usa) +
* geom_point(
*   aes(x = math, y = reading,
        color = parent_edu), 
    alpha = .7)
```

![](ggplot-concepts_files/figure-html/unnamed-chunk-17-1.png)
]
.pull-right[

```r
ggplot(pisa_usa) +
* geom_boxplot(
*   aes(x = parent_edu, y = reading,
        color = parent_edu,
        fill = parent_edu), 
    alpha = .7)
```

![](ggplot-concepts_files/figure-html/unnamed-chunk-18-1.png)
]
]

---
class: yourturn
.center[
## Your Turn
]
### Use code below to create a histogram of the math scores. &emsp;- Can you modify the width of the bins? &emsp;&emsp;(Hint: run `?geom_histogram`)

```r
ggplot(pisa_usa, aes(x = math)) + 
  ## your code here
```

### Use code below to create boxplots of math scores by sex. &emsp;- Can you make a violin plot instead a boxplot? &emsp;- Can you add color to the boxplots/violins? &emsp;&emsp;(Hint: run `?geom_violin`)

```r
ggplot(pisa_usa, aes(x = sex, y = math)) + 
  ## your code here
```

---
## Adding Layers

Begin with plot with one layer

.left-code[

```r
ggplot(
 pisa_usa, 
 aes(x = parent_edu, y = reading,
 fill = parent_edu, 
 color = parent_edu)
 ) +
* geom_boxplot(
* alpha = .2
* )
```
]
.right-plot[
<img src="ggplot-concepts_files/figure-html/unnamed-chunk-21-1.png" width="100%" />
]

---
## Adding Layers

Layers are stacked in the order of code appearance

.left-code[

```r
ggplot(
 pisa_usa, 
 aes(x = parent_edu, y = reading,
 fill = parent_edu, 
 color = parent_edu)
 ) +
 geom_boxplot( 
 alpha = .2 
 ) + 
* geom_point(
* alpha = .5
* )
```
]
.right-plot[
<img src="ggplot-concepts_files/figure-html/unnamed-chunk-22-1.png" width="100%" />
]

---
## Overwrite Global Aesthetics

.left-code[

```r
ggplot(
 pisa_usa, 
 aes(x = parent_edu, y = reading,
 fill = parent_edu, 
* color = parent_edu)
 ) +
 geom_boxplot( 
 
 alpha = .4
 ) + 
 geom_point(
* color = "black",
 alpha = .5
 ) 
```
]
.right-plot[
<img src="ggplot-concepts_files/figure-html/unnamed-chunk-23-1.png" width="100%" />
]

---
## What did we need?

### Data, Aesthetics, Geometries

### Everything else has sensible defaults

.center[
 
<img src="images/layers2.png" width="500">
]

---
class: inverse, center, middle
# Statistical Layers

---
## Statistics

### Describes how the data are modified in order to be expressed through the `geom`.

**Stats and geoms go together**.

- Every `geom` has a default `stat` and vice versa.
  - Count number of observations in each category for a bar chart
  - Calculate summary statistics for a
boxplot.

- `stat` can be specified inside of a geom and vice versa.

???

Transform input variables to displayed values:

- Count number of observations in each category for a bar chart
- Calculate summary statistics for a
boxplot.