de.lmu.ifi.dbs.elki.utilities

Class DatabaseUtil

java.lang.Object

de.lmu.ifi.dbs.elki.utilities.DatabaseUtil

public final class DatabaseUtil
extends Object

Class with Database-related utility functions such as centroid computation, covariances etc.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	DatabaseUtil.CollectionFromRelation<O> Collection view on a database that retrieves the objects when needed.
static class	DatabaseUtil.RelationObjectIterator<O> Iterator class that retrieves the given objects from the database.

Constructor Summary

Constructors

Modifier	Constructor and Description
private	DatabaseUtil() Fake constructor: Do not instantiate!

Method Summary

Methods

Modifier and Type	Method and Description
static <NV extends NumberVector<?>> Pair<NV,NV>	computeMinMax(Relation<NV> relation) Determines the minimum and maximum values in each dimension of all objects stored in the given database.
static <V extends FeatureVector<?>> int	dimensionality(Relation<V> relation) Deprecated. Use RelationUtil.dimensionality(Relation) instead!
static <V extends NumberVector<?>> double	exactMedian(Relation<V> relation, DBIDs ids, int dimension) Returns the median of a data set in the given dimension.
static <O> Class<?>	getBaseObjectClassExpensive(Relation<O> database) Do a full inspection of the database to find the base object class.
static SortedSet<ClassLabel>	getClassLabels(Database database) Retrieves all class labels within the database.
static SortedSet<ClassLabel>	getClassLabels(Relation<? extends ClassLabel> database) Retrieves all class labels within the database.
static ArrayModifiableDBIDs	getObjectsByLabelMatch(Database database, Pattern name_pattern) Find object by matching their labels.
static Relation<String>	guessLabelRepresentation(Database database) Guess a potentially label-like representation, preferring class labels.
static <O> Class<? extends O>	guessObjectClass(Relation<O> database) Do a cheap guess at the databases object class.
static Relation<String>	guessObjectLabelRepresentation(Database database) Guess a potentially object label-like representation.
static <V extends NumberVector<?>> double	quickMedian(Relation<V> relation, ArrayDBIDs ids, int dimension, int numberOfSamples) Returns the median of a data set in the given dimension by using a sampling method.
static <V extends NumberVector<?>,T extends NumberVector<?>> Relation<V>	relationUglyVectorCast(Relation<T> database) An ugly vector type cast unavoidable in some situations due to Generics.
static double[]	variances(Relation<? extends NumberVector<?>> database, NumberVector<?> centroid, DBIDs ids) Determines the variances in each dimension of the specified objects stored in the given database.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Detail

DatabaseUtil

private DatabaseUtil()

Fake constructor: Do not instantiate!

Method Detail

dimensionality

@Deprecated
public static <V extends FeatureVector<?>> int dimensionality(Relation<V> relation)

Deprecated. Use RelationUtil.dimensionality(Relation) instead!

Get the dimensionality of a relation.

Parameters:

relation - Relation

Returns:

Dimensionality

variances

public static double[] variances(Relation<? extends NumberVector<?>> database,
                 NumberVector<?> centroid,
                 DBIDs ids)

Determines the variances in each dimension of the specified objects stored in the given database.

Parameters:

database - the database storing the objects

ids - the ids of the objects

centroid - the centroid or reference vector of the ids

Returns:

the variances in each dimension of the specified objects

computeMinMax

public static <NV extends NumberVector<?>> Pair<NV,NV> computeMinMax(Relation<NV> relation)

Determines the minimum and maximum values in each dimension of all objects stored in the given database.

Type Parameters:

NV - vector type

Parameters:

relation - the database storing the objects

Returns:

Minimum and Maximum vector for the hyperrectangle

quickMedian

public static <V extends NumberVector<?>> double quickMedian(Relation<V> relation,
                                             ArrayDBIDs ids,
                                             int dimension,
                                             int numberOfSamples)

Returns the median of a data set in the given dimension by using a sampling method.

Parameters:

relation - Relation to process

ids - DBIDs to process

dimension - Dimensionality

numberOfSamples - Number of samples to draw

Returns:

Median value

exactMedian

public static <V extends NumberVector<?>> double exactMedian(Relation<V> relation,
                                             DBIDs ids,
                                             int dimension)

Returns the median of a data set in the given dimension.

Parameters:

relation - Relation to process

ids - DBIDs to process

dimension - Dimensionality

Returns:

Median value

guessLabelRepresentation

public static Relation<String> guessLabelRepresentation(Database database)
                                                 throws NoSupportedDataTypeException

Guess a potentially label-like representation, preferring class labels.

Parameters:

database -

Returns:

string representation

Throws:

NoSupportedDataTypeException

guessObjectLabelRepresentation

public static Relation<String> guessObjectLabelRepresentation(Database database)
                                                       throws NoSupportedDataTypeException

Guess a potentially object label-like representation.

Parameters:

database -

Returns:

string representation

Throws:

NoSupportedDataTypeException

getClassLabels

public static SortedSet<ClassLabel> getClassLabels(Relation<? extends ClassLabel> database)

Retrieves all class labels within the database.

Parameters:

database - the database to be scanned for class labels

Returns:

a set comprising all class labels that are currently set in the database

getClassLabels

public static SortedSet<ClassLabel> getClassLabels(Database database)

Retrieves all class labels within the database.

Parameters:

database - the database to be scanned for class labels

Returns:

a set comprising all class labels that are currently set in the database

guessObjectClass

public static <O> Class<? extends O> guessObjectClass(Relation<O> database)

Do a cheap guess at the databases object class.

Type Parameters:

O - Restriction type

Parameters:

database - Database

Returns:

Class of first object in the Database.

getBaseObjectClassExpensive

public static <O> Class<?> getBaseObjectClassExpensive(Relation<O> database)

Do a full inspection of the database to find the base object class. Note: this can be an abstract class or interface! TODO: Implement a full search for shared superclasses. But since currently the databases will always use only once class, this is not yet implemented.

Type Parameters:

O - Restriction type

Parameters:

database - Database

Returns:

Superclass of all objects in the database

getObjectsByLabelMatch

public static ArrayModifiableDBIDs getObjectsByLabelMatch(Database database,
                                          Pattern name_pattern)

Find object by matching their labels.

Parameters:

database - Database to search in

name_pattern - Name to match against class or object label

Returns:

found cluster or it throws an exception.

relationUglyVectorCast

public static <V extends NumberVector<?>,T extends NumberVector<?>> Relation<V> relationUglyVectorCast(Relation<T> database)

An ugly vector type cast unavoidable in some situations due to Generics.

Type Parameters:

V - Base vector type

T - Derived vector type (is actually V, too)

Parameters:

database - Database

Returns:

Database