2D hierarchical clustering implementation optimized for large data sets containing many small clusters i.e. dimensions of clusters << dimension of entire dataset More...

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

Inheritance diagram for GridBasedClustering< Metric >:

Public Types
typedef GridBasedCluster::Point	Point
	cluster centre, cluster bounding box, grid index More...

typedef GridBasedCluster::Rectangle	Rectangle

typedef ClusteringGrid::CellIndex	CellIndex

typedef std::multiset< MinimumDistance >::const_iterator	MultisetIterator

typedef boost::unordered::unordered_multimap< int, MultisetIterator >::const_iterator	NNIterator

Public Types inherited from ProgressLogger
enum	LogType { CMD, GUI, NONE }
	Possible log types. More...

Public Member Functions
	GridBasedClustering (Metric metric, const std::vector< double > &data_x, const std::vector< double > &data_y, const std::vector< int > &properties_A, const std::vector< int > &properties_B, std::vector< double > grid_spacing_x, std::vector< double > grid_spacing_y)
	initialises all data structures More...

	GridBasedClustering (Metric metric, const std::vector< double > &data_x, const std::vector< double > &data_y, std::vector< double > grid_spacing_x, std::vector< double > grid_spacing_y)
	initialises all data structures More...

void	cluster ()
	performs the hierarchical clustering (merges clusters until their dimension exceeds that of cell) More...

void	extendClustersY ()
	extends clusters in y-direction if possible (merges clusters further in y-direction, i.e. clusters can now span multiple cells) More...

void	removeSmallClustersY (double threshold_y)
	removes clusters with bounding box dimension in y-direction below certain threshold More...

std::map< int, GridBasedCluster >	getResults () const
	returns final results (mapping of cluster indices to clusters) More...

Public Member Functions inherited from ProgressLogger
	ProgressLogger ()
	Constructor. More...

	~ProgressLogger ()
	Destructor. More...

	ProgressLogger (const ProgressLogger &other)
	Copy constructor. More...

ProgressLogger &	operator= (const ProgressLogger &other)
	Assignment Operator. More...

void	setLogType (LogType type) const
	Sets the progress log that should be used. The default type is NONE! More...

LogType	getLogType () const
	Returns the type of progress log being used. More...

void	startProgress (SignedSize begin, SignedSize end, const String &label) const
	Initializes the progress display. More...

void	setProgress (SignedSize value) const
	Sets the current progress. More...

void	endProgress () const
	Ends the progress display. More...

Private Member Functions
void	init_ (const std::vector< double > &data_x, const std::vector< double > &data_y, const std::vector< int > &properties_A, const std::vector< int > &properties_B)
	initialises all data structures More...

bool	mergeVeto_ (const GridBasedCluster &c1, const GridBasedCluster &c2) const
	checks if two clusters can be merged Each point in a cluster can (optionally) have two properties A and B. Properties A need to be the same, properties B need to differ in each cluster. Methods checks if this is violated in the merged cluster. More...

bool	findNearestNeighbour_ (const GridBasedCluster &cluster, int cluster_index)
	determines the nearest neighbour for each cluster More...

void	eraseMinDistance_ (const std::multiset< MinimumDistance >::const_iterator it)
	remove minimum distance object and its related data More...

Private Attributes
Metric	metric_
	metric for measuring the distance between points in the 2D plane More...

ClusteringGrid	grid_
	grid on which the position of the clusters are registered used in cluster method More...

std::map< int, GridBasedCluster >	clusters_
	list of clusters maps cluster indices to clusters More...

std::map< int, GridBasedCluster >	clusters_final_
	list of final clusters i.e. clusters that are no longer merged More...

std::multiset< MinimumDistance >	distances_
	list of minimum distances stores the smallest of the distances in the head More...

boost::unordered::unordered_multimap< int, std::multiset< MinimumDistance >::const_iterator >	reverse_nns_
	reverse nearest neighbor lookup table for finding out which clusters need to be updated faster More...

boost::unordered::unordered_map< int, std::multiset< MinimumDistance >::const_iterator >	distance_it_for_cluster_idx_
	cluster index to distance iterator lookup table for finding out which clusters need to be updated faster More...

Additional Inherited Members
Static Protected Member Functions inherited from ProgressLogger
static String	logTypeToFactoryName_ (LogType type)
	Return the name of the factory product used for this log type. More...

Protected Attributes inherited from ProgressLogger
LogType	type_

time_t	last_invoke_

ProgressLoggerImpl *	current_logger_

Static Protected Attributes inherited from ProgressLogger
static int	recursion_depth_

Detailed Description

template<typename Metric>
class OpenMS::GridBasedClustering< Metric >

2D hierarchical clustering implementation optimized for large data sets containing many small clusters i.e. dimensions of clusters << dimension of entire dataset

The clustering problem therefore simplifies to a number of local clustering problems. Each of the local problems can be solved on a couple of adjacent cells on a larger grid. The grid spacing is determined by the expected typical cluster size in this region.

Each data point can have two additional properties A and B. In each cluster all properties A need to be the same, all properties B different.

Member Typedef Documentation

◆ CellIndex

typedef ClusteringGrid::CellIndex CellIndex

◆ MultisetIterator

typedef std::multiset<MinimumDistance>::const_iterator MultisetIterator

◆ NNIterator

typedef boost::unordered::unordered_multimap<int, MultisetIterator>::const_iterator NNIterator

◆ Point

typedef GridBasedCluster::Point Point

cluster centre, cluster bounding box, grid index

◆ Rectangle

typedef GridBasedCluster::Rectangle Rectangle

Constructor & Destructor Documentation

◆ GridBasedClustering() [1/2]

GridBasedClustering	(	Metric	metric,
		const std::vector< double > &	data_x,
		const std::vector< double > &	data_y,
		const std::vector< int > &	properties_A,
		const std::vector< int > &	properties_B,
		std::vector< double >	grid_spacing_x,
		std::vector< double >	grid_spacing_y
	)

inline

initialises all data structures

Parameters

data_x	x-coordinates of points to be clustered
data_y	y-coordinates of points to be clustered
properties_A	property A of points (same in each cluster)
properties_B	property B of points (different in each cluster)
grid_spacing_x	grid spacing in x-direction
grid_spacing_y	grid spacing in y-direction

◆ GridBasedClustering() [2/2]

GridBasedClustering	(	Metric	metric,
		const std::vector< double > &	data_x,
		const std::vector< double > &	data_y,
		std::vector< double >	grid_spacing_x,
		std::vector< double >	grid_spacing_y
	)

inline

initialises all data structures

Parameters

data_x	x-coordinates of points to be clustered
data_y	y-coordinates of points to be clustered
grid_spacing_x	grid spacing in x-direction
grid_spacing_y	grid spacing in y-direction

Member Function Documentation

◆ cluster()

void cluster ( )

inline

performs the hierarchical clustering (merges clusters until their dimension exceeds that of cell)

References OpenMS::Constants::c, DBoundingBox< D >::enlarge(), GridBasedCluster::getBoundingBox(), GridBasedCluster::getCentre(), GridBasedCluster::getPoints(), GridBasedCluster::getPropertiesB(), GridBasedCluster::getPropertyA(), DPosition< D, TCoordinateType >::getX(), DPosition< D, TCoordinateType >::getY(), DIntervalBase< D >::maxPosition(), and DIntervalBase< D >::minPosition().

◆ eraseMinDistance_()

void eraseMinDistance_ ( const std::multiset< MinimumDistance >::const_iterator it )

inlineprivate

remove minimum distance object and its related data

Remove the distance object behind it from distances_ and remove all corresponding data from the auxiliary data structures reverse_nns_ and distance_it_for_cluster_idx_.

Parameters

it	Iterator of distance to be removed from distances_

◆ extendClustersY()

void extendClustersY ( )

inline

extends clusters in y-direction if possible (merges clusters further in y-direction, i.e. clusters can now span multiple cells)

References ClusteringGrid::addCluster(), GridBasedCluster::getCentre(), ClusteringGrid::getClusters(), ClusteringGrid::getIndex(), ClusteringGrid::isNonEmptyCell(), and ClusteringGrid::removeCluster().

◆ findNearestNeighbour_()

bool findNearestNeighbour_	(	const GridBasedCluster &	cluster,
		int	cluster_index
	)

inlineprivate

determines the nearest neighbour for each cluster

Note: If no nearest neighbour exists, the cluster is removed from the list. The deletion is done outside of the method, see return boolean.; If two clusters cannot be merged (merge veto), they are no viable nearest neighbours.

Parameters

cluster	cluster for which the nearest neighbour should be found
cluster_index	index of cluster
Should	the cluster be removed from the cluster list?

References GridBasedCluster::getCentre(), ClusteringGrid::getClusters(), ClusteringGrid::getIndex(), and ClusteringGrid::isNonEmptyCell().

◆ getResults()

std::map<int, GridBasedCluster> getResults ( ) const

inline

returns final results (mapping of cluster indices to clusters)

◆ init_()

void init_	(	const std::vector< double > &	data_x,
		const std::vector< double > &	data_y,
		const std::vector< int > &	properties_A,
		const std::vector< int > &	properties_B
	)

inlineprivate

initialises all data structures

Parameters

data_x	x-coordinates of points to be clustered
data_y	y-coordinates of points to be clustered
properties_A	property A of points (same in each cluster)
properties_B	property B of points (different in each cluster)

References ClusteringGrid::addCluster(), GridBasedCluster::getCentre(), ClusteringGrid::getIndex(), and ClusteringGrid::removeCluster().

◆ mergeVeto_()

bool mergeVeto_	(	const GridBasedCluster &	c1,
		const GridBasedCluster &	c2
	)		const

inlineprivate

checks if two clusters can be merged Each point in a cluster can (optionally) have two properties A and B. Properties A need to be the same, properties B need to differ in each cluster. Methods checks if this is violated in the merged cluster.

Parameters

c1	cluster 1
c2	cluster 2

Returns: veto for merging clusters true -> clusters can be merged false -> clusters cannot be merged

References GridBasedCluster::getPropertiesB(), and GridBasedCluster::getPropertyA().

◆ removeSmallClustersY()

void removeSmallClustersY ( double threshold_y )

inline

removes clusters with bounding box dimension in y-direction below certain threshold

Parameters

threshold_y minimal dimension of the cluster bounding box

References DIntervalBase< D >::maxY(), and DIntervalBase< D >::minY().

Member Data Documentation

◆ clusters_

std::map<int, GridBasedCluster> clusters_

private

list of clusters maps cluster indices to clusters

◆ clusters_final_

std::map<int, GridBasedCluster> clusters_final_

private

list of final clusters i.e. clusters that are no longer merged

◆ distance_it_for_cluster_idx_

boost::unordered::unordered_map<int, std::multiset<MinimumDistance>::const_iterator> distance_it_for_cluster_idx_

private

cluster index to distance iterator lookup table for finding out which clusters need to be updated faster

◆ distances_

std::multiset<MinimumDistance> distances_

private

list of minimum distances stores the smallest of the distances in the head

◆ grid_

ClusteringGrid grid_

private

grid on which the position of the clusters are registered used in cluster method

◆ metric_

Metric metric_

private

metric for measuring the distance between points in the 2D plane

◆ reverse_nns_

boost::unordered::unordered_multimap<int, std::multiset<MinimumDistance>::const_iterator> reverse_nns_

private

reverse nearest neighbor lookup table for finding out which clusters need to be updated faster

Public Types

Public Member Functions

Private Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

template<typename Metric> class OpenMS::GridBasedClustering< Metric >

Member Typedef Documentation

◆ CellIndex

◆ MultisetIterator

◆ NNIterator

◆ Point

◆ Rectangle

Constructor & Destructor Documentation

◆ GridBasedClustering() [1/2]

◆ GridBasedClustering() [2/2]

Member Function Documentation

◆ cluster()

◆ eraseMinDistance_()

◆ extendClustersY()

◆ findNearestNeighbour_()

◆ getResults()

◆ init_()

◆ mergeVeto_()

◆ removeSmallClustersY()

Member Data Documentation

◆ clusters_

◆ clusters_final_

◆ distance_it_for_cluster_idx_

◆ distances_

◆ grid_

◆ metric_

◆ reverse_nns_

template<typename Metric>
class OpenMS::GridBasedClustering< Metric >