How to draw the following graph in R? And what are these types of graphs called?

Question

I am trying to present the following data

x <- factor(c(1,2,3,4,5))
x
[1] 1 2 3 4 5
Levels: 1 2 3 4 5
value <- c(10,5,7,4,12)
value
[1] 10  5  7  4 12
y <- data.frame(x, value)
y
  x value
1 1    10
2 2     5
3 3     7
4 4     4
5 5    12

I want to convert the above information into the following graphical representation

What is the above type of graphs called. I checked out dot plot, but that only stacks vertically.

Answer 1

This solution plots sets of three bar graphs facetted by x . The height of the bars within each set is determined using the remainder from dividing value by 3. Horizontal spacing is provided by natural geom spacing. Vertical spacing is created using white gridlines.

library(ggplot2)
library(reshape2)

Data

dataset <- data.frame('x' = 1:5, 'value' = c(10, 5, 7, 4, 12))

Since every value is supposed to be represented by three bars, we will add 3 columns to the dataset and distribute the magnitude of the value among them using integer division:

dataset[, c('col1', 'col2', 'col3')] <- floor(dataset$value / 3)
r <- dataset$value %% 3
dataset[r == 1, 'col1'] <- dataset[dataset$value %% 3 == 1, 'col1'] + 1
dataset[r == 2, c('col1', 'col2')] <- dataset[r == 2, c('col1', 'col2')] + 1

Now, we will melt the dataframe for the purposes of plotting:

dataset <- melt(dataset, id.vars = c('x', 'value'))
colnames(dataset)[4] <- 'magnitude' # avoiding colnames conflict
dataset$variable <- as.character(dataset$variable) # column ordering within a facet

Plot

First, we will make a regular bar graph. We can move facet labels to the bottom of the plot area using the switch parameter.

plt <- ggplot(data = dataset)
plt <- plt + geom_col(aes(x=variable, y = magnitude), fill = 'black')
plt <- plt + facet_grid(.~x, switch="both")

Then we will use theme_minimal() and add a few tweaks to the parameters that govern the appearance of gridlines. Specifically, we will make sure that minor XY gridlines and major X gridlines are blank, whereas major Y gridlines are white and plotted on top of the data.

plt <- plt + theme_minimal()
plt <- plt + theme(panel.grid.major.x = element_blank(),
                   panel.grid.major.y = element_line(colour = "white", size = 1.5),
                   panel.grid.minor = element_blank(),
                   panel.ontop = TRUE)

We can add value labels using geom_text() . We will only use x values from col2 records such that we're not plotting the value over each bar within each set ( col2 happens to be the middle bar).

plt <- plt + geom_text(data = dataset[dataset$variable == 'col2', ], 
                       aes(label = value, x = variable, y = magnitude + 0.5))
plt <- plt + theme(axis.text.x=element_blank()) # removing the 'col' labels
plt + xlab('x') + ylab('value')

Answer 2

The following code will do a graph similar to the one in the question.
I had to change the data.frame, yours was not fit to graph with geom_dotplot . The new variable z$value is a vector of the values 1:5 each repeated as many times as value .

library(ggplot2)

value <- c(10, 5, 7, 4, 12)

z <- sapply(value, function(v) c(1, rep(0, v - 1)))
z <- cumsum(unlist(z))
z <- data.frame(value = z)

ggplot(z, aes(x = jitter(value))) +
  geom_dotplot() +
  xlab("value")