You are here: Start » Function Reference » Clustering » KMeansClustering

KMeansClustering

Clusters data using KMeans algorithm

Syntax

void avl::KMeansClustering
(
	const avl::Matrix& inData,
	const int inClusters,
	const int inMaxIterations,
	const int inSeed,
	const float inTerminationFactor,
	const avl::KMeansClusteringMethod::Type inClusteringMethod,
	avl::Matrix& outCentroids,
	atl::Array<int>& outPointToClusterAssignment,
	float& outDistanceSum
)

void KMeansClustering
(
	Matrix inData,
	int inClusters,
	int inMaxIterations,
	int inSeed,
	float inTerminationFactor,
	KMeansClusteringMethod inClusteringMethod,
	out Matrix outCentroids,
	out int[] outPointToClusterAssignment,
	out float outDistanceSum
)

Parameters

Name	Type	Range	Default	Description
inData	const Matrix&			Data set, where variables are in columns and examples are in rows.
inClusters	const int	2 - +	2	Number of clusters to extract.
inMaxIterations	const int	10 - 1000	200	Maximal number of procedure iterations
inSeed	const int	0 -	5489	Seed to init random engine.
inTerminationFactor	const float	1.0 - 2.0	1.5f	Additional factor of procedure stop.
inClusteringMethod	const KMeansClusteringMethod::Type		KMeansPlusPlus	KMeans variant to use.
outCentroids	Matrix&			Resulting centroid points in feature space.
outPointToClusterAssignment	Array<int>&			Array of input point assignments to generated clusters.
outDistanceSum	float&			Sum of squared distances from points to its respective cluster centroids.

Errors

Error type	Description
DomainError	Empty dataset on input in KMeansClustering.
DomainError	Cannot make more clusters than there is data in input dataset.