Search code examples
rggplot2data-manipulation

Bar chart labeled with % relative to the Months in X axis

I have a data frame like this:

df <- data.frame(
     month = month.name[1:6],
     shirts = runif(6, min=1000, max=2000),
     hats = runif(6, min=1000, max=2000)
    ) |>
    pivot_longer(cols=c("shirts", "hats"), names_to="category", values_to="income")

df

The data frame:

# A tibble: 12 × 3
   month    category income
   <chr>    <chr>     <dbl>
 1 January  shirts    1782.
 2 January  hats      1489.
 3 February shirts    1094.
 4 February hats      1954.
 5 March    shirts    1467.
 6 March    hats      1483.
 7 April    shirts    1512.
 8 April    hats      1890.
 9 May      shirts    1600.
10 May      hats      1914.
11 June     shirts    1333.
12 June     hats      1609.

In order to create a bar plot with the income percentages relative for each category in each month, I did this:

df |> 
    group_by(month, category) |>
    summarize(income = sum(income, na.rm=TRUE)) |>
    ggplot(aes(x=month,y=income, fill=category)) +
    geom_bar(position="dodge", stat="identity") + 
    geom_text(aes(label=paste0(percent(income/sum(income)))), position = position_dodge(width=0.9), angle = 90, hjust=-0.1) +
    labs(x="Month",y="Income") +
    ylim(0, 4000)

But it is calculating the percentage relative to the total income sum from january to june.

I'd like this percentage to be relative to each month total income sum.

What could I change or add something new to the plot code to achieve this?

Solution

  • To use monthly totals as the denominator, you need to compute your percentages while grouped by month. The data is still grouped by month after your summarize(), so add a mutate() there:

    library(dplyr)
library(ggplot2)
library(scales)

df |> 
    group_by(month, category) |>
    summarize(income = sum(income, na.rm=TRUE)) |>
    mutate(pct_label = percent(income/sum(income), accuracy = 0.1)) |>
    ungroup() |>
    ggplot(aes(x=month,y=income, fill=category)) +
    geom_col(position="dodge") + 
    geom_text(aes(label = pct_label), position = position_dodge(width=0.9), angle = 90, hjust=-0.1) +
    labs(x="Month",y="Income") +
    ylim(0, 4000)

    Also note that geom_col() is equivalent to geom_bar(stat = "identity").