Skip to content

A Python package for creating UpSet plots using Plotly

License

Notifications You must be signed in to change notification settings

eskin22/upsetty

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

17 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

upsetty logo

Python Plotly Pandas

📖 Table of Contents

🧮 What is upsetty? Watermark

Python package for creating UpSet plots using Plotly.

🤔 What's an UpSet Plot?

An UpSet plot is a diagram used to quantitatively visualize sets and their interactions. They are particularly useful visuals for determining the overlap between different groups, as an alternative to Venn or Euler diagrams, which can become cluttered and hard to read with more than a few sets.

Comparing Venn Diagram and UpSet Plot for 5 Interacting Sets

✨ Why create upsetty?

Currently, the number of tools to create UpSet plots is very limited. Indeed, many of the previous packages for creating these plots have been deprecated or are too verbose.

To that end, we offer upsetty as a lightweight, easy-to-use alternative for analyzing overlapping sets in Python.

🤩 Like this repository? Giving a ⭐️ really helps out!

🚀 Quickstart

⬇️ Installation

pip install upsetty

⚙️ Usage

from upsetty import Upset

To create an UpSet plot, we structure the data like this:

import pandas as pd

# create sample data ({'class_name': [boolean indicators]})
data = {
    'Class A': [True, True, True, False, False, True],
    'Class B': [True, True, True, True, True, False],
    'Class C': [False, False, False, True, True, True]
}

# convert sample data dict to pd.DataFrame
df = pd.DataFrame(data)

Then, simply pass the DataFrame to the generate_plot method to create a Plotly figure of an UpSet plot.

# create UpSet figure
upset = Upset.generate_plot(df)

# show the figure
upset.show()

Using the sample data provided above, the output is pictured below:

Example UpSet Plot

Note

If you're having trouble getting the output pictured above, you can run the demo script located at upsetty/demo.py.

You can also change the colors and sizing for various aspects of the plot by passing additional parameters to the generate_plot function like so:

upset = Upset.generate_plot(
    
    # sample data
    df,

    # change category colors to a light blue, green, and yellow
    categories_colors=['#3987CA', '#FFC300', '#39CA41'],

    # change the category label color to a dark black
    categorylabel_color='#2F2F2F',

    # change the bar intersect color to a soft black
    bar_intersect_color='#454545',

    # change the marker line color to a soft black
    markerline_color='#454545'
)

Example UpSet Plot with Custom Format

By default, the function expects a DataFrame with columns of all boolean values, indicating the presence of absence of a given class.

If you wish to compute set instersection based on some other value in your data, you can do so like this:

data = {
    'Class A': [True, True, False, False, True, True, False],
    'Class B': [True, False, True, False, True, False, True],
    'Class C': [True, False, False, True, False, True, True],

    # adding a column 'Value' of non-boolean numbers
    'Value': [1, 2, 3, 4, 5, 6, 7]
}

df = pd.DataFrame(data)

upset = Upset.generate_plot(df, 'Value')
upset.show()

The code above will compute the sum of values for each subset within the classes, as depicted below.

Example UpSet Plot based on 'Value'

📌 Future Plans

Auto-adjusting margins for variable class labels

Currently, the upsetty works best with 3-4 class labels. More or less than that causes the class labels to be misaligned. Future improvements will add capabilities for auto-adjusting the margins based on the number of class labels contained in the visual.

Intersection highlighting

The ability to highlight specific intersections would give the user a way to focus their visual on specific set interactions as opposed to the basic highlighting.

Intersection count makeup

The ability to show the makeups of the different classes in a set intersection count.