How to barplot a 3 columns data frame with geom_bar?

Question

My data frame looks like this:

`cycle1 ( 14 teachers)` `cycle2 ( 16 teachers)` `cycle3 ( 11 teachers)`

1              7.14                    18.8                    27.3
2              14.3                     25                      27.3
3              21.4                     25                      18.2
4              14.3                     31.2                    36.4
5              21.4                     18.8                    45.5

and I would like to create a barchart which looks like this: barchart

I succeeded to do it with the barplot() function but I need to add confidence intervals, that's why I'm trying to do it with ggplot. The problem comes from my data's organization but I don't know how to solve it.

Please provide a [reproducible example](https://stackoverflow.com/a/5963610/8015575) — Ben373, Jun 18 '20 at 10:31

score 1 · Answer 1 · edited Jun 20 '20 at 09:12

You can use the following code

library(tidyverse)
library(viridis)


df %>% 
  pivot_longer(cols = -id) %>% 
  ggplot(aes(x = name, y=value, fill = factor(id))) + 
  geom_col(position = position_dodge()) + 
  scale_fill_viridis(discrete = T, name = "Title") + theme_bw() + 
  scale_y_continuous(limits = c(0, max(df)+1),
                     expand = c(0, 0)) + 
  labs(x="", y="Proportion (%)") +
  theme(text = element_text(family = "serif", color = "black", size = 15))+ 
  theme(axis.text = element_text(family = "serif", color = "black", size = 12))

Data

df = structure(list(id = 1:5, `cycle1 (14 teachers)` = c(7.14, 14.3, 
21.4, 14.3, 21.4), `cycle2 (16 teachers)` = c(18.8, 25, 25, 31.2, 
18.8), `cycle3 (11 teachers)` = c(27.3, 27.3, 18.2, 36.4, 45.5
)), class = "data.frame", row.names = c(NA, -5L))

yes it works thank you! I'm now trying to add the confidence intervalle — Audreygvt, Jun 18 '20 at 12:01
If it has worked, please accept this as [answer](https://stackoverflow.com/help/someone-answers). — UseR10085, Jun 18 '20 at 13:04

ljwharbers · Accepted Answer · 2020-06-18T10:59:25.083

0

ggplot takes as input data in the long format. You can use melt() to transform your data to the correct long format. For information how to use melt, you can use ?melt.

This will result in a 3 column table with in the first column the variable (your different cycles) and in the 2nd the values. Using your example data (simplifying the column names) this would result in the following code.

df = data.frame(id = as.character(1:5),
                cycle1 = c(7.14, 14.3, 21.4, 14.3, 21.4), 
                cycle2 = c(18.8, 25, 25, 31.2, 18.8), 
                cycle3 = c(27.3, 27.3, 18.2, 36.4, 45.5))

library(reshape2)
library(viridis)

df_melt = melt(df, id.vars=  "id")

ggplot(df_melt, aes(x = variable, y = value, fill = id)) +
  geom_col(position = position_dodge()) +
  scale_fill_viridis_d()

edited Jun 18 '20 at 10:59

answered Jun 18 '20 at 10:31

ljwharbers

393
2
8

I used the melt function and I get a 2 column tables but in the first one the cycles 1 for example is present in 5 columns it doesn't work with ggplot, i have this error message: Erreur : stat_count() can only have an x or y aesthetic. – Audreygvt Jun 18 '20 at 10:51
I have edited my comment with your example data. Hopefully it's clear now. – ljwharbers Jun 18 '20 at 11:00

How to barplot a 3 columns data frame with geom_bar?

2 Answers2

Data