OpenMS  2.5.0
Public Member Functions | Private Member Functions | List of all members
ClusterAnalyzer Class Reference

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

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

Public Member Functions

 ClusterAnalyzer ()
 default constructor More...
 
 ClusterAnalyzer (const ClusterAnalyzer &source)
 copy constructor More...
 
virtual ~ClusterAnalyzer ()
 destructor More...
 
std::vector< floataverageSilhouetteWidth (const std::vector< BinaryTreeNode > &tree, const DistanceMatrix< float > &original)
 Method to calculate the average silhouette widths for a clustering. More...
 
std::vector< floatdunnIndices (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< floatcohesion (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)
 Method to calculate a partition resulting from a certain step in clustering given by the number of clusters. More...
 
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

ClusterAnalyzeroperator= (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]

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_quantitydesired partition Size analogue to ClusterAnalyzer::cut
treevector of BinaryTreeNode's representing the clustering
Exceptions
invalid_parameterif 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
treevector of BinaryTreeNode's representing the clustering
originalDistanceMatrix 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
clustersvector of vectors holding the clusters (with indices to the actual elements)
originalDistanceMatrix 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_quantitySize giving the number of clusters (i.e. starting elements - cluster_quantity = cluster step)
treevector of BinaryTreeNode's representing the clustering
subtreesvector of trees holding the trees, tree is composed of cut at given size
Exceptions
invalid_parameterif 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 
)

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;

Parameters
cluster_quantitySize giving the number of clusters (i.e. starting elements - cluster_quantity = cluster step)
treevector of BinaryTreeNode's representing the clustering
clustersvector of vectors holding the clusters (with indices to the actual elements)
Exceptions
invalid_parameterif 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().

◆ 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
treevector of BinaryTreeNode's representing the clustering
originalDistanceMatrix for all clustered elements started from
tree_from_singlelinkagetrue 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
treevector of BinaryTreeNode's representing the clustering
include_distancebool value indicating whether the distance shall be included to the string
See also
BinaryTreeNode

◆ operator=()

ClusterAnalyzer& operator= ( const ClusterAnalyzer source)
private

assignment operator