How to manually fill colors in a ggplot2 histogram

I am generating a histogram and I would like to color certain groups with specific colors. Here is my histogram:

enter image description here

I have 14 groups and I would like to color the first 7 red, the next 4 blue, and the final 3 orange. How can I do this in ggplot? Thanks.

r
ggplot2

2012-04-04 17:31
by drbunsen

I'm assuming you mean a bar plot, not a histogram? There a (big) difference - joran 2012-04-04 17:36

The data is a plot of frequency from discontinuous data. I plotted it using geom_histogram. I'm not sure if this constitutes a "bar chart" or a "discrete histogram" - drbunsen 2012-04-04 17:44

Ok. I'd probably just use geom_bar in that case. And then you just need a grouping variable in your data frame that defines the color grouping you want, and then map that to fill. There are some examples in ?geom_bar - joran 2012-04-04 17:57

Thanks, I will use geom_bar. I thought there was probably a way to use geom_histogram without the need to define color groupings - drbunsen 2012-04-04 18:05

If you post a small reproducible example, we can give more specific advice - Eric Fail 2012-05-10 01:24

UPDATED VERSION

No need to specify grouping column, ggplot command is much more compact.

library(ggplot2)
set.seed(1234)

# Data generating block
df <- data.frame(x=sample(1:14, 1000, replace=T))
# Colors
colors <- c(rep("red",7), rep("blue",4), rep("orange",3))

ggplot(df, aes(x=x)) +
  geom_histogram(fill=colors) +
  scale_x_discrete(limits=1:14)

enter image description here

OLD VERSION

library(ggplot2)

# 
# Data generating block
#
df <- data.frame(x=sample(c(1:14), 1000, replace=TRUE))
df$group <- ifelse(df$x<=7, 1, ifelse(df$x<=11, 2, 3))

#
# Plotting
#
ggplot(df, aes(x=x)) +
  geom_histogram(data=subset(df,group==1), fill="red") +
  geom_histogram(data=subset(df,group==2), fill="blue") +
  geom_histogram(data=subset(df,group==3), fill="orange") +
  scale_x_discrete(breaks=df$x, labels=df$x)

enter image description here

2012-11-21 22:10
by redmode

I found the approach geom_histogram(fill=colors) to be the only way to handle a cumulative histogram aes(y=cumsum(..count../sum(..count..))), because setting a fill inside the aes resulted in groups being stacked. The simpler stat_ecdf approach didn't work for me because it doesn't take breaks option. In the end, only this approach worked - PatrickT 2017-12-18 21:36