I prefer the density plots as they depict the distributions better than histograms. Density plots are in fact smoother version of histograms.
Libraries
For today’s demonstration, we will work with the mtcars dataset. You have to know 2 variables which we are using today: mpg which is the miles per gallon variable, aka mileage, and am variable which is the transmission type: manual or automatic.
Cut me the chase and take me to the final plot or the much prettier plot.
Density plot - rough
Let’s draw a density plot on mileage and also color them by transmission type.
mtcars %>% # your data here
ggplot(aes(mpg, # inside aesthetics, add your variable of interest
fill = am
)) + # we want to fill the plots by transmission type
geom_density() # :-)
Okay and not okay. What is the problem with this figure? We did not get a colored plot as we thought of. But why? To be colored based on a group, that variable should be coded as a categorical variable, otherwise called as factor variable in R. Let’s check how am variable is coded.
Checking the structure
str(mtcars)
'data.frame': 32 obs. of 11 variables:
$ mpg : num 21 21 22.8 21.4 18.7 18.1 14.3 24.4 22.8 19.2 ...
$ cyl : num 6 6 4 6 8 6 8 4 4 6 ...
$ disp: num 160 160 108 258 360 ...
$ hp : num 110 110 93 110 175 105 245 62 95 123 ...
$ drat: num 3.9 3.9 3.85 3.08 3.15 2.76 3.21 3.69 3.92 3.92 ...
$ wt : num 2.62 2.88 2.32 3.21 3.44 ...
$ qsec: num 16.5 17 18.6 19.4 17 ...
$ vs : num 0 0 1 1 0 1 0 1 1 1 ...
$ am : num 1 1 1 0 0 0 0 0 0 0 ...
$ gear: num 4 4 4 3 3 3 3 4 4 4 ...
$ carb: num 4 4 1 1 2 1 4 2 2 4 ...Ah, we see that the am variable is coded as a numerical variable. Let’s tweak the type and recode now.
and redrawing the plot
mtcars %>%
ggplot(aes(mpg,
# note that we code am variable as a factor
fill = factor(am)
)) +
geom_density()
We got the colors based on transmission type. Overlapping regions obscure the plot ends. Hence I prefer lighter colors.
Add alpha to see overlaps
Adding alpha inside any geom adds transparency.
We see that the am group and thus the legend is coded as 0 and 1, that legends are coded so too. It would be better to spell them out, what 0 and 1 means.
To achieve this, let us recode the data using mutate command and store in a new name
Changing to factor variable
df <- mtcars %>%
mutate(
am = # am variable should be mutated (changed) to..
factor(am, # a factor variable..
levels = c(0, 1), # which has these levels
labels = c("Automatic", "Manual") # add corresponding names to the levels
)
)
# to check
str(df$am)
Factor w/ 2 levels "Automatic","Manual": 2 2 2 1 1 1 1 1 1 1 ...Yup, it is a factor with 2 levels now. Awesome!
Add title and labels
df %>%
ggplot(aes(mpg, fill = am)) +
geom_density(alpha = 0.4) +
theme_classic() +
labs( # I am adding title and axes labels using this command
x = "Miles/(US) gallon",
y = "Density",
title = "Density plots of mileage by transmission type"
)
Still the title of legends need some working; am is not intuitive.
Change legend title
df %>%
ggplot(aes(mpg, fill = am)) +
geom_density(alpha = 0.4) +
theme_classic() +
labs(
x = "Miles/(US) gallon",
y = "Density",
title = "Density plots of mileage by transmission type",
fill = "Transmission type"
)
To change the legend title, I used the fill here, as the colors fill the plots.
This is pretty good and we can end it here. But if you want to step this up, try adding the sample sizes to the density plots.
Saving sample sizes as dataframe
# sample sizes by transmission group
df %>%
group_by(am) %>%
summarize(samplesize = n())
# A tibble: 2 × 2
am samplesize
<fct> <int>
1 Automatic 19
2 Manual 13We will now use this summary from the object fortext to annotate the plots.
Prettier final plot
df %>%
ggplot(aes(mpg, fill = am)) +
geom_density(alpha = 0.8) +
geom_text( # we call this to annotate
data = fortext, # using this data we created earlier
aes(
y = y, # x and y we input manually
x = x,
label = am, # asking to label the transmission types
fontface = "bold",
color = am
)
) +
scale_fill_manual(values=c("#868E74", "#AD5988"))+
# adding sample sizes
geom_text(
data = fortext,
aes(
x = x,
y = y - 0.005, # a bit lower than the transmission type level
color = am,
fontface = "bold",
label = str_glue(
"n = {samplesize}" # I want n = before the sample sizes
)
)
) +
scale_color_manual(values=c("#868E74", "#AD5988"))+
theme_classic() +
labs(
x = "Miles/(US) gallon",
y = "Density",
title = "Density plots of mileage by transmission type"
) +
theme(legend.position = "none") # no longer required since we have labeled them in the plots
Oh how much I love it!
I would probably do this if this was my first density plot!
See you with a next one!