R: Clustered columns running totals in using ggplot2

Question

I have been trying to plot clustered columns with running totals in. I have two types of columns and just need R to calculate the running total of each them separately. For some reason it's adding running totals of different types together.

library(ggplot2)
df = data.frame(date = c(1, 1, 2, 2, 3, 3),
                val  = c(5, 2, 5, 2, 5, 2),
                type = c("honey","bread","honey","bread","honey","bread"))

ggplot(df, aes(x=date, y=cumsum(val), group = type, fill=type)) +
geom_bar(position = position_dodge(preserve = 'total'), stat="identity") + theme_classic()

I am getting:

What I am looking is to have a running total in, showing honey with values of 5,10 and 15, and bread with values of 2, 4 and 6.

What am I doing wrong? Any idea?

Answer 1

You can try something like this:

library(ggplot2)
library(dplyr)

df1 <- df %>% group_by(type) %>% mutate(N=cumsum(val))

ggplot(df1,aes(x=date, y=N, group = type, fill=type)) +
  geom_bar(position = position_dodge(preserve = 'total'), stat="identity") + theme_classic()

Which gives:

Answer 2

I would rather do the calculation within ggplot, but outside to caluclate the cumsum().

Here is my suggestion:

library(ggplot2)
library(dplyr)

df %>% 
  group_by(type) %>% 
  mutate(total = cumsum(val)) %>% 
  ggplot(aes(x = date, y = total, fill = type)) +
  geom_bar(position = position_dodge(preserve = 'total'), stat="identity") + 
  theme_classic()

... which gives you that output:

Answer 3

Another solution

ggplot(df, aes(x=date, y=ave(x = val, type, FUN = cumsum), group = type, fill=type)) +
  geom_bar(position = position_dodge(preserve = 'total'), stat="identity") + theme_classic()