Overlay histogram with empirical density and dnorm function

Question

I want to overlay a ggplot histogram (y-axis = counts) with the empirical and normal density curve. I tried:

library(ggplot2) 
set.seed(1234) 
v <- as_tibble(rnorm(1000, 10, 2.5)) 
ggplot(v, aes(x = value)) +
        geom_histogram(aes(y = ..density..), 
                       bins = 40,  colour = "black", fill = "white") +
        geom_line(aes(y = ..density.., color = 'Empirical'), stat = 'density') +     
        stat_function(fun = dnorm, aes(color = 'Normal'),
                         args = list(mean = 10, sd = 2.5)) +
        scale_colour_manual(name = "Colors", values = c("red", "blue"))

But this has the density as y scale, and I want frequencies as y-axis.

My second trial produced the plot with the frequencies (counts) as y-axis but only with the empirical density.

library(ggplot2)
set.seed(1234)
v <- as_tibble(rnorm(1000, 10, 2.5))
b  <- seq(0, 20, by = 0.5)
p1 <- ggplot(v, aes(x = value)) +
    geom_histogram(aes(y = ..count..), 
                   breaks = b,
                   binwidth = 0.5,  
                   colour = "black", 
                   fill = "white") +
    geom_line(aes(y = ..density.. * (1000 * 0.5),
                    color = 'Empirical'),
                    stat = 'density') +
    scale_colour_manual(name = "Colors", values = c("red", "blue"))

I could not manage to display also a dnorm curve in the same plot. When I tried for instance the next lines I got the density curve (blue line) on the x-axis.

p2 <- p1 + stat_function(fun = dnorm, aes(color = 'Normal'),
                     args = list(mean = 10, sd = 2.5))
p2

I assume that I have to adapt the curve with the binwidth (as with the empirical line) but I don't know how to do it.

I searched this problem in SO and could find many similar questions. But all of them addressed either my first trial (with density as y-axis), an empirical overlay with a count axis (my second trial) or used other (the base) plot commands I am not familiar with.

Actually, ggplot2 does support a 2nd y axis https://ggplot2.tidyverse.org/reference/sec_axis.html — Hack-R, Jun 26 '18 at 13:30
https://stackoverflow.com/questions/6967664/ggplot2-histogram-with-normal-curve — user20650, Jun 26 '18 at 13:32

petzi · Accepted Answer · 2018-06-26T21:07:44.937

I rewrote my code following the link from @user20650 and applied the answer by @PatrickT to my problem.

library(ggplot2)
n = 1000
mean = 10
sd = 2.5
binwidth = 0.5
set.seed(1234)
v <- as_tibble(rnorm(n, mean, sd))
b  <- seq(0, 20, by = binwidth)
ggplot(v, aes(x = value, mean = mean, sd = sd, binwidth = binwidth, n = n)) +
    geom_histogram(aes(y = ..count..), 
           breaks = b,
           binwidth = binwidth,  
           colour = "black", 
           fill = "white") +
    geom_line(aes(y = ..density.. * n * binwidth, colour = "Empirical"),
           size = 1, stat = 'density') +
    stat_function(fun = function(x) 
           {dnorm(x, mean = mean, sd = sd) * n * binwidth}, 
           aes(colour = "Normal"), size = 1) +
    labs(x = "Score", y = "Frequency") +
    scale_colour_manual(name = "Line colors", values = c("red", "blue"))

The decisive change is in the stat-function line, where the necessary adaption for n and binwidth is provided. Furthermore I did not know that one could pass parameters to aes().

Actually, I do not believe that it is a duplicate but rather a composition of different questions, prominently [here](https://stackoverflow.com/questions/27611438/density-curve-overlay-on-histogram-where-vertical-axis-is-frequency-aka-count%3E), [here](https://stackoverflow.com/questions/11404531/r-ggplot2-adding-count-labels-to-histogram-with-density-overlay) and finally [here](https://stackoverflow.com/questions/6967664/ggplot2-histogram-with-normal-curve). — petzi, Jun 26 '18 at 15:17

Overlay histogram with empirical density and dnorm function

1 Answers1

Linked