Stacked bar chart

Question

I would like to create a stacked chart using ggplot2 and geom_bar.

Here is my source data:

Rank F1     F2     F3
1    500    250    50
2    400    100    30
3    300    155    100
4    200    90     10

I want a stacked chart where x is the rank and y is the values in F1, F2, F3.

# Getting Source Data
  sample.data <- read.csv('sample.data.csv')

# Plot Chart
  c <- ggplot(sample.data, aes(x = sample.data$Rank, y = sample.data$F1))
  c + geom_bar(stat = "identity")

This is as far as i can get. I'm not sure of how I can stack the rest of the field values.

Maybe my data.frame is not in a good format?

Jurisconsult

this question is asked on a daily basis – Noni

@user2209016 Check out the documentation: docs.ggplot2.org/current/geom_bar.html. It answers a lot of common questions. – Scorpaenid

In my opinon, the link to the documentation above is a poor place to begin learning ggplot. For example, knowing that "The aesthetic mapping...only needs to be set at the layer level if you are overriding the plot defaults" is not useful to a beginner. I have found the cookbook webpages to be way more accessible. – Darkish

Answer 1

You said :

Maybe my data.frame is not in a good format?

Yes this is true. Your data is in the wide format You need to put it in the long format. Generally speaking, long format is better for variables comparison.

Using reshape2 for example , you do this using melt:

dat.m <- melt(dat,id.vars = "Rank") ## just melt(dat) should work

Then you get your barplot:

ggplot(dat.m, aes(x = Rank, y = value,fill=variable)) +
    geom_bar(stat='identity')

But using lattice and barchart smart formula notation , you don't need to reshape your data , just do this:

barchart(F1+F2+F3~Rank,data=dat)

Antimasque

Thanks, this helped along with all the other answers. Generally do you like to use lattice over ggplot2? – Jurisconsult

Good question. It short, ggeplot2 aes is unbeatable (binding data to plot features) but sometimes I find lattice panel very helpful and easier to integrate with grid package than ggplot2 – Antimasque

@CMichael No. The OP was about ggplot2, so it is normal that all the answers pointed ggplot2 not lattice. But I admit that there are more questions about ggplot2 than ``lattice` – Antimasque

Answer 2

You need to transform your data to long format and shouldn't use $ inside aes:

DF <- read.table(text="Rank F1     F2     F3
1    500    250    50
2    400    100    30
3    300    155    100
4    200    90     10", header=TRUE)

library(reshape2)
DF1 <- melt(DF, id.var="Rank")

library(ggplot2)
ggplot(DF1, aes(x = Rank, y = value, fill = variable)) + 
  geom_bar(stat = "identity")

Pervade

Answer 3

Building on Roland's answer, using tidyr to reshape the data from wide to long:

library(tidyr)
library(ggplot2)

df <- read.table(text="Rank F1     F2     F3
1    500    250    50
2    400    100    30
3    300    155    100
4    200    90     10", header=TRUE)

df %>% 
  gather(variable, value, F1:F3) %>% 
  ggplot(aes(x = Rank, y = value, fill = variable)) + 
  geom_bar(stat = "identity")

Enkindle

Answer 4

You will need to melt your dataframe to get it into the so-called long format:

require(reshape2)
sample.data.M <- melt(sample.data)

Now your field values are represented by their own rows and identified through the variable column. This can now be leveraged within the ggplot aesthetics:

require(ggplot2)
c <- ggplot(sample.data.M, aes(x = Rank, y = value, fill = variable))
c + geom_bar(stat = "identity")

Instead of stacking you may also be interested in showing multiple plots using facets:

c <- ggplot(sample.data.M, aes(x = Rank, y = value))
c + facet_wrap(~ variable) + geom_bar(stat = "identity")

Telephonist