How to avoid the points getting overlapped while using stripplot in categorical scatter plot Seaborn Library in Python?

Visualizing data is an important step since it helps understand what is going on in the data without actually looking at the numbers and performing complicated computations. Seaborn is a library that helps in visualizing data. It comes with customized themes and a high level interface.

General scatter plots, histograms, etc can't be used when the variables that need to be worked with are categorical in nature. This is when categorical scatterplots need to be used.

Plots such as 'stripplot', 'swarmplot' are used to work with categorical variables. The stripplot function is used when at least one of the variables is categorical. The data is represented in a sorted manner along one of the axes. But the disadvantage is that certain points get overlapped. This is where the jitter parameter has to be used to avoid the overlapping between variables.

Understanding the Problem

When using stripplot with categorical data, points with the same value often overlap at the exact same position, making it difficult to see the actual distribution of data. The jitter parameter adds some random noise to the dataset and adjusts the positions of the values along the categorical axis.

Syntax

seaborn.stripplot(x=None, y=None, data=None, jitter=True)

Example Without Jitter

Let's first see how overlapping occurs without using jitter ?

import pandas as pd
import seaborn as sb
import matplotlib.pyplot as plt

# Load the iris dataset
my_df = sb.load_dataset('iris')

# Create stripplot without jitter
sb.stripplot(x="species", y="petal_length", data=my_df, jitter=False)
plt.title("Stripplot without Jitter - Points Overlap")
plt.show()

Example With Jitter

Now let's see how jitter helps avoid overlapping points ?

import pandas as pd
import seaborn as sb
import matplotlib.pyplot as plt

# Load the iris dataset
my_df = sb.load_dataset('iris')

# Create stripplot with jitter
sb.stripplot(x="species", y="petal_length", data=my_df, jitter=True)
plt.title("Stripplot with Jitter - Points Spread Out")
plt.show()

Controlling Jitter Amount

You can control the amount of jitter by passing a numeric value instead of True ?

import pandas as pd
import seaborn as sb
import matplotlib.pyplot as plt

# Load the iris dataset
my_df = sb.load_dataset('iris')

# Create subplots to compare different jitter values
fig, axes = plt.subplots(1, 3, figsize=(15, 5))

# No jitter
sb.stripplot(x="species", y="petal_length", data=my_df, jitter=False, ax=axes[0])
axes[0].set_title("No Jitter (jitter=False)")

# Default jitter
sb.stripplot(x="species", y="petal_length", data=my_df, jitter=True, ax=axes[1])
axes[1].set_title("Default Jitter (jitter=True)")

# Custom jitter amount
sb.stripplot(x="species", y="petal_length", data=my_df, jitter=0.3, ax=axes[2])
axes[2].set_title("Custom Jitter (jitter=0.3)")

plt.tight_layout()
plt.show()

Key Parameters

Alternative: Using Swarmplot

For better visualization without overlapping, you can also use swarmplot which automatically adjusts point positions ?

import pandas as pd
import seaborn as sb
import matplotlib.pyplot as plt

# Load the iris dataset
my_df = sb.load_dataset('iris')

# Create swarmplot as alternative
sb.swarmplot(x="species", y="petal_length", data=my_df)
plt.title("Swarmplot - No Overlapping Points")
plt.show()

Conclusion

The jitter parameter in stripplot effectively prevents point overlapping by adding random noise along the categorical axis. Use jitter=True for default spacing or specify a numeric value to control the amount of jitter. For small datasets, consider using swarmplot as an alternative.