Library to run the indidually-fair clustering algorithms of the paper

Scalable Individually-Fair K-Means Clustering, AISTATS 2024

Usage

First, create a fresh virtual environment and install the requirements.

# From google_research/
virtualenv -p python3 .
source ./bin/activate

python3 -m pip install -r individually_fair_clustering/requirements.txt

Then, run the algorithm. Example command:

python3 -m individually_fair_clustering.run_individually_fair_clustering
--input=path-to-input.tsv
--output=path-to-output.json
--k=10
--algorithm="LSPP"

The input of consists of a file in tab separated format with each row being a point and each column being a dimension of the point. All dimensions are float. All points should have the same dimension.

The output is a json file in text format encoding a data frame with a single row. The data frame contains statistics about the result of the algorithm (e.g., the k-means cost).