R training: How to make graphs

1.1 plot()

The quickest way to plot data in R is by using the plot() function. Simply call a column in your dataset and wrap the function around it:

# to make the graphs less crowded we pick the first 100 entries from the dataset
data2 <- slice_head(data, n = 100)

# now we plot the sliced data
plot(data2$calc_total_income)

When you create plot like this, it appears in the “Plots” pane on the bottom right. Every circle stands for one observation.

You can also plot two variables against each other to study the relationship between them. The following example shows that income and expenditure are somewhat positively correlated.

plot(data2$calc_total_income, data2$calc_total_expenditure)

1.2 barplot()

A simple bar plot with frequencies of responses can be made like this:

# first, create an object displaying the frequencies of answer options
freq <- table(data$nationality)

# then plot the data
barplot(freq)

1.3 hist()

To better understand your dataset, a quick histogram can be quite handy:

hist(data2$calc_total_income)

1.4 boxplot()

A quick boxplot can be drawn as simple as this:

boxplot(data2$calc_total_expenditure)

back to top

2.1 Basic syntax

ggplot2 always follows the same basic syntax. First, you call the ggplot() function, which typically includes an argument calling the data frame with the values you want to plot, and an aes() statement, in which you specify the different dimension of your plot. Additional arguments are then added to the ggplot() function as needed using the + operator, which is the ggplot2 equivalent of dplyr’s pipe operator (%>%).

A basic bar plot displaying the counts per nationality is built by adding geo_bar() to the expression like so:

ggplot(data, aes(x = nationality)) + 
  geom_bar()

You could make this graph look nicer by adding a couple of elements to the expression:

ggplot(data, aes(x = nationality, fill = nationality)) + 
  geom_bar() +
  scale_fill_manual(values = c(rgb(0.93,0.35,0.35), rgb(0.97,0.67,0.67),
                               rgb(0.35,0.35,0.35), rgb(0.67,0.67,0.68))) +
  labs(title = "Number of surveys per nationality", fill = "Nationality") +
  theme(plot.title = element_text(face= "bold", size = 12, hjust = 0.5),
        legend.position = "none",
        panel.background = element_blank(),
        axis.title.x = element_blank(),
        axis.title.y = element_blank(),
        panel.grid.major.y = element_line(colour = "gainsboro"),
        axis.ticks = element_blank()) +
  scale_x_discrete(labels=c("iran" = "Iranian", "palestinian_territories" = "Palestinian",
                              "syria" = "Syrian", "turkey" = "Turkish"))

There are many ways to style your plots. You can use pre-defined themes as well as change individual elements. Whatever it is that you want to modify, google is your friend.

2.2 Long format

In order to plot data with ggplot2, you will need your data to be in long format for some cases (check training “3. Intermediate: Additional R functions” for more details). Let’s say we wanted to create boxplots for income and expenditure (in the same plot). We would transform our data frame like this:

data_long <- gather(data2, variable, value, c(calc_total_income, calc_total_expenditure))

Then we can proceed to plot the data like so:

boxplot <- ggplot(data = data_long, aes(x = variable, y = value)) +
  geom_boxplot()

boxplot

The same logic with long data applies to line graphs with multiple groups. Lets say we wanted to plot the number of daily interviews per nationality:

# First, we make sure the today column has date properties.
data$today <- as.Date(data$today, format = "%d/%m/%Y")

# Then, we summarize the data by counting the number of observations per nationality and day.
data_day <- data %>%
  group_by(nationality, today) %>%
  summarize(obs = n())

# Let's have a look at the data frame with the summarized data.
head(data_day)

## # A tibble: 6 x 3
## # Groups:   nationality [1]
##   nationality today        obs
##   <chr>       <date>     <int>
## 1 iran        2020-09-01     6
## 2 iran        2020-09-02     5
## 3 iran        2020-09-03     5
## 4 iran        2020-09-04    12
## 5 iran        2020-09-05     6
## 6 iran        2020-09-07    24

# The summarize function already returns an output in long format, so we can directly start plotting.
## Note that the aes() expression now has more arguments (i.e. "group") than in the basic example above,
## which is only possible with data in long format.
line <- data_day %>%
  ggplot(aes(x = today, y = obs, group = nationality, color = nationality)) +
  geom_line() + theme(legend.position="bottom")

line

back to top