Create stacked bar chart of within group totals

Question

I have the dataset below:

eg_data <- data.frame(
time = c("1","2"), 
size1=c(200, 300))

I add a row to the dataset that is a sum of both time periods:

eg_data <- rbind(eg_data, (data.frame(time="1 + 2", 
size1=sum(eg_data$size1))))
eg_data$time <- factor(eg_data$time, levels = c("1 + 2", "1", "2"))

I create a bar graph

eg_data_bar <- (ggplot() +
geom_bar(data=eg_data, aes(y = size1, x = time, fill = time), 
stat="identity",position='stack'))

eg_data_bar

As it stands now, the graph renders out as three bars which are different colors. I need the bar chart to render as three bars, period 1 is one color, period 2 is another color, and the grouped period 1 + 2 is made up of the two colors from period 1 and period 2.

The problem I'm having is how to define fill within the geom_bar(). When I've done stacked bars before, I always have a third column to use as the fill.

Any help is appreciated.

"and the grouped period 1 + 2 is made up of the two colors from period 1 and period 2" I'm not sure I understand. What do you mean by that? Are you after some form of pattern fill, showing two colours for the same bar? As far as I know that will be impossible (difficult?) in ggplot2. — Maurits Evers
I need the final output to be three bars...period 1, period 2, and period 1 + 2. I need period 1 to be...red, period 2 to be...blue, and period 1 + 2 to period 1 and period 2, stacked on top of each other, so you can see the relationship b/t p1 totals, p2 totals, and how p1 and p2 look as proportions of the overall total. In a single visual. — Adam_S

Maurits Evers Maurits Evers · Accepted Answer · 2018-12-06T23:17:28

Perhaps something like this?

We start from your initial eg_data data.frame:

eg_data <- data.frame(
    time = c("1","2"),
    size1 = c(200, 300))


library(tidyverse)
eg_data %>%
    mutate(grp = factor(time, levels = 1:2)) %>%
    group_by(time) %>%
    complete(grp) %>%
    replace_na(list(size1 = 0)) %>%
    ungroup() %>%
    spread(time, size1) %>%
    mutate(`1+2` = `1` + `2`) %>%
    gather(time, size, -grp) %>%
    mutate(time = factor(time, levels = c("1", "2", "1+2"))) %>%
    ggplot(aes(time, size, fill = grp)) +
    geom_col(position = "stack")

The key is to add an additional grp variable which we use to fill the bars.

Solution below is solution above, it adds grp to the dataframe and creates the graph as its own object, so it can be called ad-hoc.

eg_data <- eg_data %>% 
mutate(grp = factor(time, levels = 1:2)) %>%
group_by(time) %>%
complete(grp) %>%
replace_na(list(size1 = 0)) %>%
ungroup() %>%
spread(time, size1) %>%
mutate(`1+2` = `1` + `2`) %>%
gather(time, size1, -grp) %>%
mutate(time = factor(time, levels = c("1", "2", "1+2")))

eg_data_graph <-  (ggplot(data = eg_data, aes(time, size1, fill = grp)) +
geom_col(position = "stack"))

eg_data_graph

Create stacked bar chart of within group totals

1 Answers