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

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

Public Member Functions
	ClusterAnalyzer ()
	default constructor More...

	ClusterAnalyzer (const ClusterAnalyzer &source)
	copy constructor More...

virtual	~ClusterAnalyzer ()
	destructor More...

std::vector< float >	averageSilhouetteWidth (const std::vector< BinaryTreeNode > &tree, const DistanceMatrix< float > &original)
	Method to calculate the average silhouette widths for a clustering. More...

std::vector< float >	dunnIndices (const std::vector< BinaryTreeNode > &tree, const DistanceMatrix< float > &original, const bool tree_from_singlelinkage=false)
	Method to calculate Dunns indices for a clustering. More...

std::vector< float >	cohesion (const std::vector< std::vector< Size > > &clusters, const DistanceMatrix< float > &original)
	Method to calculate the cohesions of a certain partition. More...

float	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. More...

void	cut (const Size cluster_quantity, const std::vector< BinaryTreeNode > &tree, std::vector< std::vector< Size > > &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. More...

String	newickTree (const std::vector< BinaryTreeNode > &tree, const bool include_distance=false)
	Returns the hierarchy described by a clustering tree as Newick-String. More...

Private Member Functions
ClusterAnalyzer &	operator= (const ClusterAnalyzer &source)
	assignment operator More...

Detailed Description

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

Constructor & Destructor Documentation

◆ ClusterAnalyzer() [1/2]

ClusterAnalyzer ( )

default constructor

◆ ClusterAnalyzer() [2/2]

ClusterAnalyzer ( const ClusterAnalyzer & source )

copy constructor

◆ ~ClusterAnalyzer()

virtual ~ClusterAnalyzer ( )

virtual

destructor

Member Function Documentation

◆ averagePopulationAberration()

float 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

◆ averageSilhouetteWidth()

std::vector<float> averageSilhouetteWidth	(	const std::vector< BinaryTreeNode > &	tree,
		const DistanceMatrix< float > &	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 cluster step

The average silhouette width will be calculated 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

◆ cohesion()

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

◆ cut() [1/2]

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

◆ cut() [2/2]

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

@brief 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 subtract that from the number of leaves, to get the number of clusters formed , i.e. cluster_quantity = data.size() - real_leaf_count;

@param cluster_quantity Size giving the number of clusters (i.e. starting elements - cluster_quantity = cluster step)
@param tree vector of BinaryTreeNode's representing the clustering
@param clusters vector of vectors holding the clusters (with indices to the actual elements)
@throw invalid_parameter if desired clusterstep is invalid
@see BinaryTreeNode

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

Referenced by SpectraMerger::mergeSpectraPrecursors().

◆ dunnIndices()

std::vector<float> dunnIndices	(	const std::vector< BinaryTreeNode > &	tree,
		const DistanceMatrix< float > &	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

◆ newickTree()

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

◆ operator=()

ClusterAnalyzer& operator= ( const ClusterAnalyzer & source )

private

assignment operator

Public Member Functions

Private Member Functions

Detailed Description

Constructor & Destructor Documentation

◆ ClusterAnalyzer() [1/2]

◆ ClusterAnalyzer() [2/2]

◆ ~ClusterAnalyzer()

Member Function Documentation

◆ averagePopulationAberration()

◆ averageSilhouetteWidth()

◆ cohesion()

◆ cut() [1/2]

◆ cut() [2/2]

◆ dunnIndices()

◆ newickTree()

◆ operator=()