Histograms with variable size binwidth in ggplot2

Question

I have the following data

 dati <- read.table(text="
        class     num
    1     0.0   63530
    2     2.5   27061
    3     3.5   29938
    4     4.5   33076
    5     5.6   45759
    6     6.5   72794
    7     8.0  153177
    8    10.8  362124
    9    13.5  551051
    10   15.5  198634
  ")

And I want to produce a histogram with variable size bins, so that the area of each bar reflects the total numerosity (num) of each bin. I tried

bins <- c(0,4,8,11,16)
p <- ggplot(dati) +
  geom_histogram(aes(x=class,weight=num),breaks = bins)

however, this produces a histogram where the length of each bar is equal to total numerosity of each bin. Because bin widths are variable, areas are not proportional to numerosity. I could not solve this apparently easy problem within ggplot2. Can anyone help me?

Community · Accepted Answer · 2017-05-23 12:26:22Z

6

I think you are looking for a density plot - this closely related question has most of the answer. You call y = ..density.. in geom_histogram().

This works because stat_bin (recall geom_histogram() is geom_bar() + stat_bin(), and stat_bin() constructs a data frame with columns count and density. Thus calling y = ..density.. pulls the right column for density, whereas the default (counts) is as if you call y = ..count...

##OP's code
ggplot(dati) +  geom_histogram(aes(x=class, weight=num),
 breaks = bins)

Count Histogram

##new code (density plot)
ggplot(dati) +  geom_histogram( aes(x=class,y = ..density.., weight=num),
 breaks = bins, position = "identity")

Density Histogram

You can find some further examples in the online ggplot2 help page for geom_histogram().

edited May 23, 2017 at 12:26

CommunityBot

11 silver badge

answered Oct 15, 2013 at 19:52

Blue Magister

13.5k5 gold badges40 silver badges58 bronze badges

Sign up to request clarification or add additional context in comments.

3 Comments

Massimo2013 Over a year ago

Thanks a lot. This is more or less what I was looking for. However, because I need to compare different populations, I would like that the total area equal the total numerosity of the population rather than one. Is it possible to scale the histogram?

Blue Magister Over a year ago

Possibly by setting y = ..density.. * sum(..count..), but I hesitate to give that answer because it sounds like there may be a better way to do what you're looking for. A new question illustrating your desired output might give you a better method.

Massimo2013 Over a year ago

Thanks for your help. I posted a new question, in which I explain in detail what I need to do

Community · Accepted Answer · 2017-05-23 12:10:30Z

1

It sounds to me like you asking for how to produce variable sized bar widths. If so, you just need to call the 'width' parameter in your ggplot aesthetics like this:

ggplot(data, aes(x = x, y = y, width = num))

this method is discussed more in the following question: Variable width bars in ggplot2 barplot in R

edited May 23, 2017 at 12:10

CommunityBot

11 silver badge

answered Oct 15, 2013 at 20:50

slizb

6,2304 gold badges28 silver badges23 bronze badges

Collectives™ on Stack Overflow

Histograms with variable size binwidth in ggplot2

2 Answers 2

3 Comments

Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

3 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related