Data Visualization for Political Scientists

class: center, middle, inverse, title-slide

# Data Visualization for Political Scientists
## Session 2 - The Anatomy of a ggplot2 plot
<html>
<div style="float:left">

</div>
<hr color='#EB811B' size=1px width=796px>
</html>
### Felix Haass
### 15 Januar 2018

---

# ggplot2

The "gg" in `ggplot2` stands for the "Grammar of Graphics." The grammar of graphics is a philosophy of data visualization which forces you to think about *what* you want to visualize *how*. [Hadley Wickham](http://hadley.nz/) followed this philosophy to implement the `ggplot2` package.

---

# The anatomy of a ggplot2 plot

The grammar of graphics specifies **building blocks** out of which an analyst builds a plot. These include, in the order of application:

1. Data (*what* do you want to plot?)
2. Aesthetic mapping (what comes on the x and y axes? )
3. Geometric object (`geoms`)  (How do we want to see our data? Points, lines, bars, ...)
4. Add more `geoms` (e.g. add regression lines to a scatterplot)
5. Polish labels, scales, legends, and appearance.

(see [this link](http://socviz.co/makeplot.html#build-your-plots-layer-by-layers) for more details)

---

class: inverse
background-image: url("Ninja-header.svg_opacity1.png")
background-size: contain

# Useful tips from the dataviz ninja

1. **Think hard about *what* you want to visualize!**

> "Think of graphs as comparison" - [Andrew Gelman](http://andrewgelman.com/2014/03/25/statistical-graphics-course-statistical-graphics-advice/)

---

# ggplot2 building blocks

Let's look at the ggplot2 building blocks in practice:

```r
library(gapminder) # loads the gapminder data
library(tidyverse) # loads ggplot2 and other packages

example_plot <- ggplot(data = gapminder, # specify which dataset to use
                       aes(x = year, # what goes on the x axis?
                           y = lifeExp )) + # what's on the y axis?
  geom_point() # with which geometric object should the data be displayed?
```

Note the `+` that ties the building blocks together.

---

# ggplot2 building blocks

```r
print(example_plot)
```

---

# Aesthetics - Size

```r
library(gapminder) 
library(tidyverse)

example_plot <- ggplot(data = gapminder, 
                       aes(x = year,  # the aes() function defines aesthetics
                           y = lifeExp, 
                           size = gdpPercap)) + # map the aesthetic 'size' to gdp/pc
  geom_point()

# print(example_plot)
```

---

# Aesthetics - Size

```r
print(example_plot)
```

---

# Aesthetics II - Color

```r
library(gapminder) 
library(tidyverse)

example_plot <- ggplot(data = gapminder, 
                       # the aes() function defines aesthetics
                       aes(x = year,  # x axis
                           y = lifeExp, # y axis
                           color = continent, # map color to continent
                           size = gdpPercap)) + # map the aesthetic 'size' to gdp/pc
  geom_point() 
```

---

# Aesthetics II - Color

```r
print(example_plot)
```

---

class: inverse
background-image: url("Ninja-header.svg_opacity1.png")
background-size: contain

# Useful tips from the dataviz ninja

1. Think hard about *what* you want to visualize!

2. **Don't use too many aesthetics - just use those that help you clarify your comparison!**
   > "When ggplot successfully makes a plot but the result looks insane, the reason is almost always that something has gone wrong in the mapping between the data and aesthetics for the geom being used" - [Kieran Healy](http://socviz.co/groupfacettx.html#grouped-data-and-the-group-aesthetic)

---

# geoms

```r
library(gapminder) 
library(tidyverse)

example_plot <- ggplot(data = gapminder, 
                       aes(x = year, 
                           y = lifeExp)) +
  geom_line() # lines instead of points
```

---

# geoms

Whoops! What happened here?

```r
print(example_plot)
```

---

# geoms

```r
library(gapminder) 
library(tidyverse)

example_plot <- ggplot(data = gapminder, 
                       aes(x = year, 
                           y = lifeExp, 
                           group = country)) +  # tell ggplot2 which 
                                                # observations belong together
  geom_line() 
```

---

# geoms

```r
print(example_plot)
```

---

# Combining geoms

```r
library(gapminder) 
library(tidyverse)

example_plot <- ggplot(data = gapminder, 
                       aes(x = year, 
                           y = lifeExp)) +  
  geom_point() +
  geom_smooth(method = "lm") # add regression line
```

---

# Combining geoms

```r
print(example_plot)
```

---

# Combining geoms II

```r
library(gapminder) 
library(tidyverse)

example_plot <- ggplot(data = gapminder, 
                       aes(x = year, 
                           y = lifeExp)) +  
  geom_point() +
  geom_smooth(method = "lm") +
  geom_smooth(method = "loess", 
              color = "firebrick") # fix smoother color
```

Bonus question: in this example we fix the color, i.e. we map it to a fixed value (`firebrick` which is red). What happens if we would map `color` to a variable in the gapminder dataset, such as `continent`?

---

# Combining geoms II

```r
print(example_plot)
```

---

# Manipulate and Preprocess Data

Subsetting/filtering data helps to reduce complexity & get at the comparison that we want. To do that, we use the `dplyr` package which is part of the `tidyverse`.

To filter data, we use the `filter()` function.

```r
library(tidyverse) # loads dplyr package, among others
library(gapminder)

gapminder_americas <- gapminder %>% # the %>% `chains` together functions
  filter(continent == "Americas") # that's two "="

head(gapminder_americas, 5)
```

```
## # A tibble: 5 x 6
##     country continent  year lifeExp      pop gdpPercap
##      <fctr>    <fctr> <int>   <dbl>    <int>     <dbl>
## 1 Argentina  Americas  1952  62.485 17876956  5911.315
## 2 Argentina  Americas  1957  64.399 19610538  6856.856
## 3 Argentina  Americas  1962  65.142 21283783  7133.166
## 4 Argentina  Americas  1967  65.634 22934225  8052.953
## 5 Argentina  Americas  1972  67.065 24779799  9443.039
```

---

# Manipulate and Preprocess Data

Modify/add variables to existing data frame. We modify data with the `mutate()` function and chain them together using the pipe operator ` %>% `.

```r
library(tidyverse) # loads dplyr package, among others
library(gapminder)

gapminder_americas <- gapminder %>% 
  filter(continent == "Americas") %>% 
  # create a character/categorical variable
  # to distinguish between North/South America
  mutate(north_america = ifelse(country == "United States" |
                                  country == "Canada", 
                                "north_america", 
                                "south_america"))
head(gapminder_americas,3)
```

```
## # A tibble: 3 x 7
##     country continent  year lifeExp      pop gdpPercap north_america
##      <fctr>    <fctr> <int>   <dbl>    <int>     <dbl>         <chr>
## 1 Argentina  Americas  1952  62.485 17876956  5911.315 south_america
## 2 Argentina  Americas  1957  64.399 19610538  6856.856 south_america
## 3 Argentina  Americas  1962  65.142 21283783  7133.166 south_america
```

---

# Manipulate and Preprocess Data

Use filtered and preprocessed data to highlight comparisons in ggplot:

```r
ggplot(gapminder_americas, # only use data for Americas
       aes(x = year, 
           y = gdpPercap, 
           color = north_america)) + # map "north_america" category to color
  geom_point()
```

---

# Exercise

Plot the development of population size (`pop` variable in the gapminder data) over time (`year` variable in the gapminder data) in Asia (hint: `continent == "Asia"`). Add a trend line and/or smooth line.

Bonus exercise: Plot the relationship between population size `pop` and `gdpPercap`! (hint: might make sense to wrap `pop` and `gdpPercap` in `log()`).

---

# Solution

```r
library(tidyverse)
library(gapminder)

gapminder_asia <- gapminder %>% 
  filter(continent == "Asia")

asia_pop <- ggplot(gapminder_asia, 
                   aes(x = year, y = pop)) +
  geom_point() +
  geom_smooth(method = "lm")

print(asia_pop)
```

---

# Walkthrough Exercise

Goal:

---

# What do we want to visualize?

Think about the data! What is the comparison?

Genocide vs. non-genocide countries => Rwanda vs. rest of Africa

```r
library(gapminder)
library(tidyverse)

gapminder_africa <- gapminder %>% 
  # filter only African countries
  filter(continent == "Africa") %>% 
  # create a categorical variable that distinguishes
  # between Rwanda and other African countries
  mutate(color_plot = ifelse(country != "Rwanda", # != = "!" + "="
                             "Other African Countries", 
                             "Rwanda"))
```

---

# Add geom_line() + map color/alpha

```r
rwanda_plot <- ggplot(gapminder_africa, 
                      aes(x = year, 
                          y = lifeExp, 
                          group = country, 
                          color = color_plot)) + 
  geom_line(aes(alpha = color_plot)) # map alpha to "color_plot" variable     
                                     # ggplot chooses alpha level automatically

print(rwanda_plot)
```

---

# Add color/alpha scales

```r
rwanda_plot <- ggplot(gapminder_africa, 
                      aes(x = year, 
                          y = lifeExp, 
                          group = country, 
                          color = color_plot)) + 
  geom_line(aes(alpha = color_plot)) + 
  
  # we assign colors/alpha values/other "aes" through "scale" functions
  scale_alpha_discrete("", range = c(0.5, 1)) +  
  scale_color_manual("", values = c("lightgrey", "black"))

print(rwanda_plot)
```

---

# Manipulate appearance: add theme

```r
rwanda_plot <- ggplot(gapminder_africa, 
                      aes(x = year, 
                          y = lifeExp, 
                          group = country, 
                          color = color_plot)) + 
  geom_line(aes(alpha = color_plot)) + 
  scale_alpha_discrete("", range = c(0.5, 1)) +  
  scale_color_manual("", values = c("lightgrey", "black")) + 
  
  # add theme
  theme_bw() + # black and white theme
  theme(legend.position = "bottom", # legend position
        panel.grid = element_blank()) # remove grid lines
```

---

# Manipulate appearance: add theme

```r
print(rwanda_plot)
```

---

# Manipulate appearance: change labels

```r
rwanda_plot <- ggplot(gapminder_africa, 
                      aes(x = year, 
                          y = lifeExp, 
                          group = country, 
                          color = color_plot)) + 
  geom_line(aes(alpha = color_plot)) + 
  scale_alpha_discrete("", range = c(0.5, 1)) +  
  scale_color_manual("", values = c("lightgrey", "black")) + 
  theme_bw() + 
  theme(legend.position = "bottom", 
        panel.grid = element_blank()) +
  
  # labels, captions, and title/subtitle
  labs(x = "", y = "Life Expectancy in Years", 
       title = "The Impact of Genocide on Life Expectancy", 
       subtitle = "Life expectancy for newborns extrapolated from mortality rate in a given year.", 
       caption = " Data source: gapminder.org")
```

---

# Manipulate appearance: change labels

```r
print(rwanda_plot)
```

---

class: inverse
background-image: url("Ninja-header.svg_opacity1.png")
background-size: contain

# Useful tips from the dataviz ninja

1. Think hard about *what* you want to visualize!

2. Don't use too many aesthetics - just use those that help you clarify your comparison!

3. **Trial and error is your friend!**   
    > "If you are unsure of what each piece of code does, take advantage of ggplot's additive character. Working backwards from the bottom up, remove each + some_function(...) statement one at a time to see how the plot changes."  - [Kieran Healy](http://socviz.co/groupfacettx.html#facet-to-make-small-multiples)