A density plot visualises the distribution of data over a continuous interval (or time period). Density Plots are not affected by the number of bins (each bar used in a typical histogram) used, thus, they are better at visualizing the shape of the distribution than a histogram unless the bins in the histogram have a theoretical meaning.
- Notice that the variable on the x-axis should be continuous. Density plots are not designed for use with discrete variables.
# You may need to install seaborn on the command line using 'pip install seaborn' or 'conda install seaborn' import seaborn as sns # Set a theme for seaborn sns.set_theme(style="darkgrid") # Load the example diamonds dataset diamonds = sns.load_dataset("diamonds") # Take a look at the data print(diamonds.head())
sns.kdeplot(data=diamonds, x="price", cut=0);
This is basic, but there are lots of ways to adjust it through keyword arguments (you can see these by running
help(sns.kdeplot)) or via calling functions on the matplotlib
ax object that running
sns.kdeplot returns when not followed by
;. In this simple example, the
cut keyword argument forces the density estimate to end at the end-points of the data–which makes sense for a variable like price, which has a hard cut-off at 0.
Let’s use further keyword arguments to enrich the plot, including different colours (‘hues’) for each cut of diamond. One keyword argument that may not be obvious is
hue_order. The default function call would have arranged the
cut types so that the ‘Fair’ cut obscured the other types, so the argument passed to the
hue_order keyword below reverses the order of the unique list of diamond cuts via
sns.kdeplot(data=diamonds, x="price", hue="cut", hue_order=diamonds['cut'].unique()[::-1], fill=True, alpha=.4, linewidth=0.5, cut=0.);
For this R demonstration, we are going to use ggplot2 package to create a density plot. Additionally, we will use the dataset
diamonds that is made available by the ggplot2 package.
To begin with this R demonstration, make sure that we install and load all the useful packages that we need it.
# load necessary packages library(ggplot2) library(viridis) library(RColorBrewer) library(tidyverse) library(ggthemes) library(ggpubr) library(datasets)
Next, in order to make a density plot, we are going to use the
geom_density() functions. We will specify
price as our x-axis.
ggplot(diamonds, aes(x = price)) + geom_density()
We can always change the color of the density plot using the
col argument and fill the color inside the density plot using
fill argument. Furthermore, we can specify the degree of transparency density fill area using the argument
alpha ranges from 0 to 1.
ggplot(diamonds, aes(x = price))+ geom_density(fill = "lightblue", col = 'black', alpha = 0.6)
We can also change the type of line of the density plot as well by adding
ggplot(diamonds, aes(x = price)) + geom_density(fill = "lightblue", col = 'black', linetype = "dashed")
Furthermore, you can also combine both histogram and density plots together.
ggplot(diamonds, aes(x = price)) + geom_histogram(aes(y = ..density..), colour = "black", fill = "grey45") + geom_density(col = "red", size = 1,linetype = "dashed")
What happen if we want to make multiple densities?
For example, we want to make multiple densities plots for price based on the type of cut, all we need to do is adding
ggplot(data=diamonds, aes(x = price, fill = cut)) + geom_density(adjust = 1.5, alpha = .3)
For this demonstration, we will use the
plotplainblind scheme, a community-contributed color and grpah scheme for plots that greatly improves over Stata’s default plot color schemes especially for colorblind viewers. For more on using schemes in Stata, see here.
* Install the blindschemes set of graph schemes, including plottig ssc install blindschemes * Shows the set of available schemes graph query, schemes * Load diamonds data import delimited "https://vincentarelbundock.github.io/Rdatasets/csv/ggplot2/diamonds.csv", clear
We can build a basic density plot using the
kdensity subcommand of
* Plot the kernel density with plotplain theming twoway kdensity price, scheme(plotplainblind)
To overlay densities for multiple variables or multiple groups, it is possible to use the standard
twoway graph stacking syntax:
* Plot the density of two separate columns twoway (kdensity depth) (kdensity table), scheme(plotplainblind) * Plot the same variable separately by group, overlaid on a single set of axes twoway (kdensity price if cut == "Fair", lcolor(blue)) (kdensity price if cut == "Good", lcolor(red)), scheme(plotplainblind) legend(lab(1 "Cut: Fair") lab(2 "Cut: Good"))
However, the syntax for doing separate densities by group can get onerous very quickly with more than a handful of groups, noting that you’ll have to specify each group with an
if by hand, be careful about the color/presentation of each line, and do the legend yourself.
Much easier for by-group kernel densities is the mkdensity package, which still uses
kdensity under the hood, but just handles some of this busywork for you. On the other hand it doesn’t accept a
scheme() option. But you can still use it via
The downside of this approach, rather than doing it by hand, is that it relies on
set scheme plotplainblind * if necessary, install with ssc install mkdensity mkdensity price, over(cut)