icx 1.0.2

This package provides the functionality for an interactive Streamlit dashboard designed to support stakeholders in exploring individual fairness notions within algorithmic decision-making systems.

Individual Consistency eXplorer (icx) Python Package

This package provides the functionality for an interactive Streamlit dashboard designed to support stakeholders in exploring individual fairness notions within algorithmic decision-making systems.

The dashboard allows users to:

• Explore and operate on a tabular dataset of individuals provided with their corresponding binary classifications;
• Define how similarity between individuals is measured, by configuring categorisation of attributes and how distances between attribute values are computed;
• Compute and visualise five individual fairness metrics that summarise the consistency of classifications across the dataset; and
• Inspect attributes of specific individuals and of those individuals most similar to them, to explore variations in attribute values and allow like-for-like comparisons of classifications.

This package implements the functionality described in a ECAI Demo Track paper 2025, available at Individual Consistency eXplorer (ICX): An Interactive Dashboard for the Exploration of Individual Fairness

To see an online version of the dashboard, see Individual Consistency eXplorer Online.

To see an online version of the dashboard, see Individual Consistency eXplorer Online.

📦 Installation

It is recommended to install icx in a virtual environment (e.g., conda).

pip install icx

Basic Usage

from icx import dashboard

Run the dashboard

dashboard.run()

In the dashboard there is the ability to upload your own datasets or use the demo datasets provided.

Shield:

This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.