The Grammar of Graphics

Answers

library(tidyverse)
library(magrittr)
library(gapminder)
data(gapminder)

1. Slide 38

ggplot(gapminder, aes(y=lifeExp, x=gdpPercap, col=continent))+geom_point(alpha=0.5)

Figure 1: Plot of life Expectancy versus GDP per capita

It’s impossible to see what is going on in Figure 1! Now lets look at the same graph with both axes scaled by factors of 10 (in other words, on logarithmic axes).

p <- ggplot(gapminder, aes(y=log(lifeExp), x=log(gdpPercap), col=continent))+geom_point(alpha=0.5)
p

Figure 2: Plot of log of life Expectancy versus log of GDP per capita

In Figure 2, the points are spread pretty evenly along both axes - so you can really see what’s going on.

According to Figure 1 and Figure 2 countries with higher GDP tend to have a higher life expectancy. According to Figure 2 we can clearly see African countries and most of the Asian countries have much lower life expectancy and GDP per capita.

When does one use the logarithmic scale to visualize data instead of the absolute values?

• When the dataset contains numbers of very different magnitudes, you should consider whether a log transform will enhance the visualization. A log transformation preserves the order of the observations while making outliers less extreme.

Extra: Make ggplot2 Graphics Interactive with plotly

library(plotly)
ggplotly(p)

2. Slide 63

gapminder2007 <- gapminder %>%
  filter(year == 2007)
ggplot(gapminder2007, 
aes(x=lifeExp, col=continent, fill=continent))+
  geom_density(alpha=0.5) 

3. Slide 66

ggplot(gapminder2007, 
aes(x=continent, fill=continent))+ 
 geom_bar() 

4. Slide 73

gapminder %>%
filter(country == "India") %>%
ggplot(aes(x = year, y = gdpPercap)) +
geom_line() + 
  geom_point() + 
  labs(title="Time series plot of GDP per capita",
       x="Year", 
       y="GDP per capita (USD, inflation-adjusted)")

5. Slide 75

avglifeExp <- gapminder %>%
  group_by(continent, year) %>%
  summarise(meanlifeExp=mean(lifeExp))

ggplot(avglifeExp, aes(x=year, y=meanlifeExp, col=continent))+
  geom_line() + geom_point()

6. Slide 76

gapminder %>%
  filter(year %in% c(1952, 1957, 1962, 1967, 1972, 1977)) %>%
  filter(continent %in% c("Asia", "Americas")) %>%
  ggplot(aes(y=lifeExp, x=gdpPercap, color=continent)) + 
  geom_point() + 
  facet_wrap(~year, ncol=3)+
  labs(title="Life Expectancy vs GDP - America and Asia",
       y = "Life Expectancy",
       x = "GDP per capita")

7. Slide 77

gapminder %>%
  filter(year %in% c(1952, 1957, 1962, 1967, 1972, 1977)) %>%
  filter(continent %in% c("Asia", "Americas")) %>%
  ggplot(aes(y=log(lifeExp), x=log(gdpPercap), color=continent)) + 
  geom_point() + 
  facet_wrap(~year, ncol=3)+
  labs(title="Life Expectancy vs GDP - America and Asia",
       y = "log(Life Expectancy)",
       x = "log(GDP per capita)")

8. Slide 79

ggplot(gapminder_unfiltered, aes(gdpPercap, lifeExp, color = year)) + 
geom_point() +
facet_wrap(~ continent)

9. Slide 82

ggplot(gapminder, aes(y=lifeExp,  x=year)) + 
geom_smooth() +
facet_wrap(~ continent)

10. Slide 83

ggplot(gapminder, aes(y=lifeExp,  x=year)) + 
geom_smooth() +
  geom_point() + 
facet_wrap(~ continent)

11. Slide 84

xn <- seq(-5, 5, length=10000)
yn <- dnorm(xn)
df <- data.frame(x=xn, y=yn)
ggplot(df, aes(x=xn, y=yn))+
  geom_line(col="red")

Note:

You should include Figure captions for all graphs. It is important to interpret all graphs.