CSSS 508, Week 2

We'll be working with data from Hans Rosling's Gapminder project. An excerpt of this data can be accessed through an R package called gapminder, cleaned and assembled by Jenny Bryan at UBC.

In the console: install.packages("gapminder") (have you set your default repository yet?)

Load the package:

library(gapminder)

The data frame we will work with is called gapminder, available once you have loaded the package. Let's see its structure:

str(gapminder)

Classes 'tbl_df', 'tbl' and 'data.frame':   1704 obs. of  6 variables:
 $ country  : Factor w/ 142 levels "Afghanistan",..: 1 1 1 1 1 1 1 1 1 1 ...
 $ continent: Factor w/ 5 levels "Africa","Americas",..: 3 3 3 3 3 3 3 3 3 3 ...
 $ year     : int  1952 1957 1962 1967 1972 1977 1982 1987 1992 1997 ...
 $ lifeExp  : num  28.8 30.3 32 34 36.1 ...
 $ pop      : int  8425333 9240934 10267083 11537966 13079460 14880372 12881816 13867957 16317921 22227415 ...
 $ gdpPercap: num  779 821 853 836 740 ...

• Factor variables country and continent: categorical data with an underlying numeric representation
• Many observations! \( n=1704 \) rows
• A nested/hierarchical structure: year in country in continent

We'll want to be able to slice up this data frame into subsets (e.g. just the rows for Afghanistan, just the rows for 1997). We will use a package called dplyr to do this neatly.

In the console: install.packages("dplyr")

dplyr is a very useful and powerful package that we will talk more about soon, but today we're just going to use it for “filtering” data.

library(dplyr)

Attaching package: 'dplyr'

The following objects are masked from 'package:stats':

    filter, lag

The following objects are masked from 'package:base':

    intersect, setdiff, setequal, union

When you load packages in R that have functions sharing the same name as functions you already have, the more recently loaded package trumps the previous ones (“masks”).

This message is just letting you know that. To avoid this in your R Markdown file, add message=FALSE to your chunk options when loading packages.

Sometimes you may get a warning message, often if you aren't running the latest version of R:

Warning message:
package ‘gapminder’ was built under R version 3.2.3 

Chunk option warning=FALSE will hide this.

gapminder %>% filter(country == "Oman")

Source: local data frame [12 x 6]

   country continent  year lifeExp     pop gdpPercap
    (fctr)    (fctr) (int)   (dbl)   (int)     (dbl)
1     Oman      Asia  1952  37.578  507833  1828.230
2     Oman      Asia  1957  40.080  561977  2242.747
3     Oman      Asia  1962  43.165  628164  2924.638
4     Oman      Asia  1967  46.988  714775  4720.943
5     Oman      Asia  1972  52.143  829050 10618.039
6     Oman      Asia  1977  57.367 1004533 11848.344
7     Oman      Asia  1982  62.728 1301048 12954.791
8     Oman      Asia  1987  67.734 1593882 18115.223
9     Oman      Asia  1992  71.197 1915208 18616.707
10    Oman      Asia  1997  72.499 2283635 19702.056
11    Oman      Asia  2002  74.193 2713462 19774.837
12    Oman      Asia  2007  75.640 3204897 22316.193

We used == for testing “equals”: country == "Oman". We can also use other operators:

• !=: not equal to
• >, >=, <, <=: less than, less than or equal to, etc.
• %in%: used with checking equal to one of several values

Or we can combine multiple logical conditions:

• &: both conditions need to hold (AND)
• |: at least one condition needs to hold (OR)
• !: negates a logical condition (TRUE -> FALSE, FALSE -> TRUE)

gapminder %>%
    filter( (country == "Oman") &
                (year > 1980) & (year <= 2000) )

Source: local data frame [4 x 6]

  country continent  year lifeExp     pop gdpPercap
   (fctr)    (fctr) (int)   (dbl)   (int)     (dbl)
1    Oman      Asia  1982  62.728 1301048  12954.79
2    Oman      Asia  1987  67.734 1593882  18115.22
3    Oman      Asia  1992  71.197 1915208  18616.71
4    Oman      Asia  1997  72.499 2283635  19702.06

If we think a particular subset will be used repeatedly, we can save it and give it a name:

China <- gapminder %>% filter(country == "China")
head(China, 4)

Source: local data frame [4 x 6]

  country continent  year  lifeExp       pop gdpPercap
   (fctr)    (fctr) (int)    (dbl)     (int)     (dbl)
1   China      Asia  1952 44.00000 556263527  400.4486
2   China      Asia  1957 50.54896 637408000  575.9870
3   China      Asia  1962 44.50136 665770000  487.6740
4   China      Asia  1967 58.38112 754550000  612.7057

An alternative way of plotting many prefer (myself included) – though this is not without debate – uses the ggplot2 package in R. Install it now, and then load it.

library(ggplot2)

The core idea underlying this package is the grammar of graphics: we can break up elements of a plot into pieces and combine them.

ggplot(data = China, aes(x = year, y = lifeExp)) +
    geom_point()

ggplot(data = China, aes(x = year, y = lifeExp)) +
    geom_point(color = "red", size = 5) +
    xlab("Year") + ylab("Life expectancy") +
    ggtitle("Life expectancy in China") +
    theme_bw()

ggplot(data = gapminder, aes(x = year, y = lifeExp, color = continent)) +
    geom_point() +
    xlab("Year") + ylab("Life expectancy") +
    ggtitle("Life expectancy over time") +
    theme_bw()

ggplot(data = gapminder, aes(x = year, y = lifeExp, group = country, color = continent)) +
    geom_line(alpha = 0.5) +
    facet_wrap( ~ continent) +
    xlab("Year") + ylab("Life expectancy") +
    ggtitle("Life expectancy over time") +
    theme_bw()

Layers are the components of the graph, such as:

• ggplot(): initializes ggplot2 object, specifies input data
• geom_point(): layer of scatterplot points
• geom_line(): layer of lines
• ggtitle(), xlab(), ylab(): layer of labels
• facet_wrap(): layer creating separate panels stratified by some factor wrapping around
• facet_grid(): same idea, but can split by two variables along rows and columns (e.g. facet_grid(gender ~ age_group))
• theme_bw(): replace default gray background with black-and-white

Layers are separated by a + sign. For clarity, I put each layer on a new line.

Aesthetics control the appearance of the layers:

• x, y: \( x \)- and \( y \)-values to use
• color: set color of elements based on some data value
• group: describe which points are conceptually grouped together for the plot (often used with lines)
• size: set size of points/lines based on some data value

Layers take arguments to control their appearance, such as point/line colors or transparency (alpha between 0 and 1).

• Arguments like color, size, linetype, shape, fill, and alpha can be used directly on the layers (setting aesthetics), e.g. geom_point(color = "red"). See the ggplot2 documentation for options. These don't depend on the data.

• Arguments inside aes() (mapping aesthetics) will depend on the data, e.g. geom_point(aes(color = continent)).

• aes() in the ggplot() layer gives overall aesthetics to use in other layers, but can be changed on individual layers (including switching x or y to different variables)

(Pedantic, but searching for help will be easier for you.)

We can assign a ggplot object to a name:

lifeExp_by_year <- ggplot(data = gapminder,
       aes(x = year, y = lifeExp, color = continent)) +
    geom_point() +
    xlab("Year") +
    ylab("Life expectancy") +
    ggtitle("Life expectancy over time") +
    theme_bw()

The graph won't be displayed when you do this. You can show the graph using a single line of code with just the object name, or take the object and add more layers.

lifeExp_by_year

lifeExp_by_year +
    geom_line(aes(group = country))

Start experimenting with making some graphs in ggplot2 of the Gapminder data. You can look at a subset of the data using filter to limit rows, plot different \( x \) and \( y \) variables, facet by a factor, etc.

Some other options:

• geom_histogram(), geom_density(), geom_boxplot() (see the Cookbook for R site for a reference)
• geom_smooth() for adding loess or regression lines (see the ggplot2 documentation)
• Install Jeff Arnold's ggthemes package, load it, and try theme_economist(), theme_stata(), theme_excel() instead of no theme or theme_bw()

ggplot(data = gapminder, aes(x = continent, y = year, color = continent)) +
    geom_point()

ggplot(data = gapminder, aes(x = continent, y = year, color = continent)) +
    geom_point(position = position_jitter(width = 0.5, height = 2))

We can modify the axes in a variety of ways, such as:

• Change the \( x \) or \( y \) range using xlim() or ylim() layers
• Change to a logarithmic or square-root scale on either axis: scale_x_log10(), scale_y_sqrt()
• Change where the major/minor breaks are: scale_x_continuous(breaks =, minor_breaks = )

ggplot(data = China, aes(x = year, y = gdpPercap)) +
    geom_line() +
    scale_y_log10(breaks = c(1000, 2000, 3000, 4000, 5000)) +
    xlim(1940, 2010) +
    ggtitle("Chinese GDP per capita")

ggplot(data = China, aes(x = year, y = lifeExp)) +
    geom_line() +
    ggtitle("Chinese life expectancy") +
    theme_gray(base_size = 30)

Text size, labels, tick marks, etc. can be messed with more precisely using arguments to the theme() layer. Note: theme() is a different layer than theme_gray() or theme_bw(), which you might also be using in a previous layer. See the ggplot2 documentation for details.

Examples:

• plot.title = element_text(size = rel(2), hjust = 0) makes the title twice as big as usual and left-aligns it
• axis.text.x = element_text(angle = 45) rotates \( x \) axis labels
• axis.text = element_text(colour = "blue") makes the \( x \) and \( y \) axis labels blue
• axis.ticks.length = unit(.5, "cm") makes the axis ticks longer

Scales control how the mapped aesthetics appear. You can modify these with a scale_[aesthetic]_[option]() layer where [aesthetic] is color, shape, linetype, alpha, size, fill, etc. and [option] is something like manual, continuous or discrete (depending on nature of the variable).

Examples:

• scale_linetype_manual(): manually specify the linetype for each different value
• scale_alpha_continuous(): varies transparency over a continuous range
• scale_color_brewer(palette = "Spectral"): uses a palette from http://colorbrewer2.org (great site for picking nice plot colors!)

When confused…Google it!

lifeExp_by_year +
    scale_color_manual(name = "Which\ncontinent\nare we\nlooking at?", values = c("Africa" = "seagreen", "Americas" = "turquoise1", "Asia" = "royalblue", "Europe" = "violetred1", "Oceania" = "yellow"))

ggplot(data = gapminder, aes(x = year, y = lifeExp, group = country)) +
    geom_line(alpha = 0.5, aes(color = "Country", size = "Country")) +
    geom_line(stat = "smooth", method = "loess", aes(group = continent, color = "Continent", size = "Continent"), alpha = 0.5) +
    facet_wrap(~ continent, nrow = 2) +
    scale_color_manual(name = "Unit", values = c("Country" = "black", "Continent" = "dodgerblue1")) +
    scale_size_manual(name = "Unit", values = c("Country" = 1, "Continent" = 3)) +
    theme_minimal(base_size = 16) + 
    theme(legend.position=c(0.75, 0.2))

Observation: one could use filter to identify the countries with dips in life expectancy and investigate.

You can move the legends around, flip their orientation, remove them altogether, etc. The Cookbook for R website is my go-to for burning questions such as how to change the legend labels.

When you knit an R Markdown file, any plots you make are automatically saved in a folder in .png format. If you want to save another copy (perhaps of a different file type for use in a manuscript), use ggsave():

ggsave("I_saved_a_file.pdf", plot = lifeExp_by_year,
       height = 3, width = 5, units = "in")

If you didn't manually set font sizes, these will usually come out at a reasonable size given the dimensions of your output file.

Bad/non-reproducible way: choose Export on the plot preview.

To really wow everyone, install the gganimate package, which adds a frame aesthetic.

Note: gganimate isn't available on CRAN, so needs to be installed from github using the devtools package. Uncomment the lines to do this for the first time.

# install.packages("devtools")
# devtools::install_github("dgrtwo/gganimate")
library(gganimate)

Note: this requires the tool ImageMagick to be installed for converting to a .gif.

gdp_vs_life <- ggplot(gapminder,
                      aes(x = gdpPercap,
                          y = lifeExp,
                          size = pop,
                          color = continent,
                          frame = year)) +
    geom_point() +
    scale_x_log10() +
    theme_bw()
gg_animate(gdp_vs_life, "Lectures/animated.gif",
           ani.height = 600, ani.width = 900)

Pick some relationship to look at in the Gapminder data and write up a .Rmd file investigating that question graphically. You might work with a subset of the data (e.g. just Africa). Upload both the .Rmd file and the .html file to Canvas. Include 4 to 8 graphs. All titles, axes, and legends should be labelled clearly (no raw variable names). You must have at least one graph with facet_wrap or facet_grid. You must include at least one manually specified legend. You can use other geoms like histograms, bar charts, add vertical or horizontal lines, etc.

Your document should be pleasant for a peer to look at, with some organization. You must write up your observations in words as well as showing the graphs. Use chunk options echo and results to limit the code/output you show in the .html.

• 0: Didn't turn anything in
• 1: Turned in but low effort, ignoring many directions
• 2: Decent effort, followed directions with some minor issues
• 3: Nailed it