Plotting with categorical data

In seaborn, there are several different ways to visualize a relationship involving categorical data. Similar to the relationship between relplot() and either scatterplot() or lineplot(), there are two ways to make these plots. There are a number of axes-level functions for plotting categorical data in different ways and a figure-level interface, catplot(), that gives unified higher-level access to them.

These families represent the data using different levels of granularity. When deciding which to use, you’ll have to think about the question that you want to answer. The unified API makes it easy to switch between different kinds and see your data from several perspectives.

In this tutorial, we’ll mostly focus on the figure-level interface, catplot(). Remember that this function is a higher-level interface each of the functions above, so we’ll reference them when we show each kind of plot, keeping the more verbose kind-specific API documentation at hand.

The default representation of the data in catplot() uses a scatterplot. There are actually two different categorical scatter plots in seaborn. They take different approaches to resolving the main challenge in representing categorical data with a scatter plot, which is that all of the points belonging to one category would fall on the same position along the axis corresponding to the categorical variable. The approach used by stripplot(), which is the default “kind” in catplot() is to adjust the positions of points on the categorical axis with a small amount of random “jitter”:

The second approach adjusts the points along the categorical axis using an algorithm that prevents them from overlapping. It can give a better representation of the distribution of observations, although it only works well for relatively small datasets. This kind of plot is sometimes called a “beeswarm” and is drawn in seaborn by swarmplot(), which is activated by setting kind="swarm" in catplot():

Unlike with numerical data, it is not always obvious how to order the levels of the categorical variable along its axis. In general, the seaborn categorical plotting functions try to infer the order of categories from the data. If your data have a pandas Categorical datatype, then the default order of the categories can be set there. If the variable passed to the categorical axis looks numerical, the levels will be sorted. But the data are still treated as categorical and drawn at ordinal positions on the categorical axes (specifically, at 0, 1, …) even when numbers are used to label them:

We’ve referred to the idea of “categorical axis”. In these examples, that’s always corresponded to the horizontal axis. But it’s often helpful to put the categorical variable on the vertical axis (particularly when the category names are relatively long or there are many categories). To do this, swap the assignment of variables to axes:

As the size of the dataset grows, categorical scatter plots become limited in the information they can provide about the distribution of values within each category. When this happens, there are several approaches for summarizing the distributional information in ways that facilitate easy comparisons across the category levels.

This behavior is called “dodging” and is turned on by default because it is assumed that the semantic variable is nested within the main categorical variable. If that’s not the case, you can disable the dodging:

이와 연관된 기능인 박슨플롯boxenplot()은 (마땅한 번역어가 없어서 그냥 씁니다) 박스 플롯과 비슷한 도표를 그리기는 하지만, 그 분포의 "모양"을 좀 더 전달하는데 최적화되어 있다. 좀 더 큰 데이터셋에게 적합한 기능이다.

A different approach is a violinplot(), which combines a boxplot with the kernel density estimation procedure described in the distributions tutorial:

It can also be useful to combine swarmplot() or striplot() with a box plot or violin plot to show each observation along with a summary of the distribution:

A special case for the bar plot is when you want to show the number of observations in each category rather than computing a statistic for a second variable. This is similar to a histogram over a categorical, rather than quantitative, variable. In seaborn, it’s easy to do so with the countplot() function:

An alternative style for visualizing the same information is offered by the pointplot() function. This function also encodes the value of the estimate with height on the other axis, but rather than showing a full bar, it plots the point estimate and confidence interval. Additionally, pointplot() connects points from the same hue category. This makes it easy to see how the main relationship is changing as a function of the hue semantic, because your eyes are quite good at picking up on differences of slopes

Additionally, the axes-level functions accept vectors of Pandas or numpy objects rather than variables in a DataFrame:

To control the size and shape of plots made by the functions discussed above, you must set up the figure yourself using matplotlib commands:

This is the approach you should take when you need a categorical figure to happily coexist in a more complex figure with other kinds of plots.