ClusterAnalyzer Class Reference [Clustering]

Bundles analyzing tools for a clustering (given as sequence of BinaryTreeNode's). More...

#include <OpenMS/COMPARISON/CLUSTERING/ClusterAnalyzer.h>

List of all members.

Public Member Functions
	ClusterAnalyzer ()
	default constructor
	ClusterAnalyzer (const ClusterAnalyzer &source)
	copy constructor
virtual	~ClusterAnalyzer ()
	destructor
std::vector< Real >	averageSilhouetteWidth (const std::vector< BinaryTreeNode > &tree, const DistanceMatrix< Real > &original)
	Method to calculate the average silhouette widths for a clustering.
std::vector< Real >	dunnIndices (const std::vector< BinaryTreeNode > &tree, const DistanceMatrix< Real > &original, const bool tree_from_singlelinkage=false)
	Method to calculate Dunns indices for a clustering.
std::vector< Real >	cohesion (const std::vector< std::vector< Size > > &clusters, const DistanceMatrix< Real > &original)
	Method to calculate the cohesions of a certain partition.
Real	averagePopulationAberration (Size cluster_quantity, std::vector< BinaryTreeNode > &tree)
	Method to calculate the average aberration from average population in partition resulting from a certain step in clustering.
void	cut (const Size cluster_quantity, const std::vector< BinaryTreeNode > &tree, std::vector< std::vector< Size > > &clusters)
	Method to calculate a partition resulting from a certain step in clustering given by the number of clusters.
void	cut (const Size cluster_quantity, const std::vector< BinaryTreeNode > &tree, std::vector< std::vector< BinaryTreeNode > > &subtrees)
	Method to calculate subtrees from a given tree resulting from a certain step in clustering given by the number of clusters.
String	newickTree (const std::vector< BinaryTreeNode > &tree, const bool include_distance=false)
	Returns the hierarchy described by a clustering tree as Newick-String.
Private Member Functions
ClusterAnalyzer &	operator= (const ClusterAnalyzer &source)
	assignment operator

---

Detailed Description

Bundles analyzing tools for a clustering (given as sequence of BinaryTreeNode's).

---

Constructor & Destructor Documentation

ClusterAnalyzer

(

)

default constructor

ClusterAnalyzer

(

const ClusterAnalyzer &

source

)

copy constructor

virtual ~ClusterAnalyzer

(

)

[virtual]

destructor

---

Member Function Documentation

Real averagePopulationAberration	(	Size	cluster_quantity,
		std::vector< BinaryTreeNode > &	tree
	)

Method to calculate the average aberration from average population in partition resulting from a certain step in clustering.

Parameters:

	cluster_quantity	desired partition Size analogue to ClusterAnalyzer::cut
	tree	vector of BinaryTreeNode's representing the clustering

Exceptions:

invalid_parameter

if desired clustering is invalid

Returns:

the average aberration from the average cluster population (number of elements/cluster_quantity) at cluster_quantity

See also:

BinaryTreeNode

std::vector<Real> averageSilhouetteWidth	(	const std::vector< BinaryTreeNode > &	tree,
		const DistanceMatrix< Real > &	original
	)

Method to calculate the average silhouette widths for a clustering.

Parameters:

	tree	vector of BinaryTreeNode's representing the clustering
	original	DistanceMatrix for all clustered elements started from

Returns:

a vector filled with the average silhouette widths for each clusterstep

The average silhouette width will becalculated for each clustering step beginning with the first step(n-1 cluster) ending with the last (1 cluster, average silhouette width is 0 by definition).

See also:

BinaryTreeNode

std::vector<Real> cohesion	(	const std::vector< std::vector< Size > > &	clusters,
		const DistanceMatrix< Real > &	original
	)

Method to calculate the cohesions of a certain partition.

Parameters:

	clusters	vector of vectors holding the clusters (with indices to the actual elements)
	original	DistanceMatrix for all clustered elements started from

Returns:

a vector that holds the cohesions of each cluster given with clusters (order corresponds to clusters)

void cut	(	const Size	cluster_quantity,
		const std::vector< BinaryTreeNode > &	tree,
		std::vector< std::vector< Size > > &	clusters
	)

Method to calculate a partition resulting from a certain step in clustering given by the number of clusters.

If you want to fetch all clusters which were created with a threshold, you simply count the number of tree-nodes which are not -1, and substract that from the number of leafes, to get the number of clusters formed , i.e. cluster_quantity = data.size() - real_leaf_count;

Parameters:

	cluster_quantity	Size giving the number of clusters (i.e. starting elements - cluster_quantity = cluster step)
	tree	vector of BinaryTreeNode's representing the clustering
	clusters	vector of vectors holding the clusters (with indices to the actual elements)

Exceptions:

invalid_parameter

if desired clusterstep is invalid

See also:

BinaryTreeNode

after call of this method the argument clusters is filled corresponding to the given cluster_quantity with the indices of the elements clustered

Referenced by SpectraMerger::mergeSpectraPrecursors().

void cut	(	const Size	cluster_quantity,
		const std::vector< BinaryTreeNode > &	tree,
		std::vector< std::vector< BinaryTreeNode > > &	subtrees
	)

Method to calculate subtrees from a given tree resulting from a certain step in clustering given by the number of clusters.

Parameters:

	cluster_quantity	Size giving the number of clusters (i.e. starting elements - cluster_quantity = cluster step)
	tree	vector of BinaryTreeNode's representing the clustering
	subtrees	vector of trees holding the trees, tree is composed of cut at given size

Exceptions:

invalid_parameter

if desired clusterstep is invalid

See also:

BinaryTreeNode

after call of this method the argument clusters is filled corresponding to the given cluster_quantity with the indices of the elements clustered

std::vector<Real> dunnIndices	(	const std::vector< BinaryTreeNode > &	tree,
		const DistanceMatrix< Real > &	original,
		const bool	tree_from_singlelinkage = false
	)

Method to calculate Dunns indices for a clustering.

Parameters:

	tree	vector of BinaryTreeNode's representing the clustering
	original	DistanceMatrix for all clustered elements started from
	tree_from_singlelinkage	true if tree was created by SingleLinkage, i.e. the distances are the minimal distances in increasing order and can be used to speed up the calculation

See also:

BinaryTreeNode

String newickTree	(	const std::vector< BinaryTreeNode > &	tree,
		const bool	include_distance = false
	)

Returns the hierarchy described by a clustering tree as Newick-String.

Parameters:

	tree	vector of BinaryTreeNode's representing the clustering
	include_distance	bool value indicating whether the distance shall be included to the string

See also:

BinaryTreeNode

ClusterAnalyzer& operator=

(

const ClusterAnalyzer &

source

)

[private]

assignment operator