A refined approach for choosing initial points for k-means clustering. More...

#include <refined_start.hpp>

Public Member Functions
	RefinedStart (const size_t samplings=100, const double percentage=0.02)
	Create the RefinedStart object, optionally specifying parameters for the number of samplings to perform and the percentage of the dataset to use in each sampling.

template<typename MatType >
void	Cluster (const MatType &data, const size_t clusters, arma::mat &centroids) const
	Partition the given dataset into the given number of clusters according to the random sampling scheme outlined in Bradley and Fayyad's paper, and return centroids. More...

template<typename MatType >
void	Cluster (const MatType &data, const size_t clusters, arma::Row< size_t > &assignments) const
	Partition the given dataset into the given number of clusters according to the random sampling scheme outlined in Bradley and Fayyad's paper, and return point assignments. More...

size_t	Samplings () const
	Get the number of samplings that will be performed.

size_t &	Samplings ()
	Modify the number of samplings that will be performed.

double	Percentage () const
	Get the percentage of the data used by each subsampling.

double &	Percentage ()
	Modify the percentage of the data used by each subsampling.

template<typename Archive >
void	serialize (Archive &ar, const uint32_t)
	Serialize the object.

Detailed Description

A refined approach for choosing initial points for k-means clustering.

This approach runs k-means several times on random subsets of the data, and then clusters those solutions to select refined initial cluster assignments. It is an implementation of the following paper:

@inproceedings{bradley1998refining,
  title={Refining initial points for k-means clustering},
  author={Bradley, Paul S and Fayyad, Usama M},
  booktitle={Proceedings of the Fifteenth International Conference on Machine
      Learning (ICML 1998)},
  volume={66},
  year={1998}
}

Member Function Documentation

◆ Cluster() [1/2]

template<typename MatType >

void mlpack::kmeans::RefinedStart::Cluster	(	const MatType &	data,
		const size_t	clusters,
		arma::mat &	centroids
	)		const

Partition the given dataset into the given number of clusters according to the random sampling scheme outlined in Bradley and Fayyad's paper, and return centroids.

Partition the given dataset according to Bradley and Fayyad's algorithm.

Template Parameters

MatType Type of data (arma::mat or arma::sp_mat).

Parameters

data	Dataset to partition.
clusters	Number of clusters to split dataset into.
centroids	Matrix to store centroids into.

◆ Cluster() [2/2]