GuipengLi / Dcluster

Clustering by fast search and find of density peaks

Clustering by fast search and find of density peaks

This Python package implements the clustering algorithm proposed by Alex Rodriguez and Alessandro Laio. It generates the initial rho and delta values for each observation then use these values to assign observations to clusters.

Installation

This version is for both python2 and python3. The first step is to install Python. Python is available from the Python project page. Dcluster depends on numpy and matplotlib. The next step is to install Dcluster.

You can download the source code at Github or at PyPi for Dcluster, and then run:

$ python setup.py install

Or install from PyPi using pip , a package manager for Python:

$ pip install Dcluster

Usage

The only input is the distance metrics between observations. See the test.dat. Dcluster supports interacive clustering based on Decision Graph:

import Dcluster as dcl
filein="test.dat"
dcl.run(fi=filein, sep='\t')

Test data

See the test.dat in test/. One can choose different cluster centers based on Decision Graph. And please first press key 'n' then 'Enter' to quit. Result will be saved automatically.

Contact

Author: Guipeng Li

Email: guipeng.lee(AT)gmail.com

References

Rodriguez, A., & Laio, A. (2014). Clustering by fast search and find of density peaks. Science, 344(6191), 1492-1496. (paper)