Greedy-K-means

Greedy K-means Spark Package in Python Version 1.0, 2016-July-29

Copyright (c) 2016 Hongfu Liu & Nikos Vlassis.

1. Introduction =============== GKCC is a Python package for Greedy K-means in the Spark environment. Reference: [1] Likas, A., Vlassis, N., & Verbeek, J. J. (2003). The global k-means clustering algorithm. Pattern recognition, 36(2), 451-461.

2. Installation and Basic Usage =============================== A. Copy this file to the specific directory in your workspace.

B. Start Spark and go to your Spark path

C. Submit the job

./bin/spark-submit YOUR PATH/GKmeans.py YOUR DATA NUM_CLUSTER

3. Functions in the package ===========================

   • main: we run Greedy K-means from 1 center to K centers.
   • parseVector: transforms the data into an array.
   • closestPoint: finds the nearest center.
   • CandidateCenter: adds one center.
   • BasicKmeans: runs the basic K-means.

4. Contact ========== For questions and comments, please feel free to contact Hongfu Liu(liu.hongf@husky.neu.edu).