This repository contains codes and examples for UCSL : Unsupervised Clustering driven by Supervised Learning (UCSL) algorithm.
We aim at unsupervisedly identify subtypes and drive it with a supervised task (classification or regression).
The algorithm is implemented in a pedagogical, easy-to-use sklearn fashion with fit(), fit_predict(), predict() and predict_clusters() functions.
Method codes are available in "ucsl" directory, examples implemented in notebook are implemented in "examples" directory.
Unless you already have sklearn, Numpy and Scipy installed, you need to install them:
sudo apt-get install python-numpy python-scipy
Clone the repository from github
git clone https://github.com/rlouiset/UCSL.git
Add UCSL in your $PYTHONPATH
import numpy as np