Package elki.utilities.datastructures.histogram

Class AbstractObjDynamicHistogram<T>

java.lang.Object

elki.utilities.datastructures.histogram.AbstractStaticHistogram

elki.utilities.datastructures.histogram.ObjHistogram<T>

elki.utilities.datastructures.histogram.AbstractObjDynamicHistogram<T>

Type Parameters:

T - Data type

All Implemented Interfaces:

Histogram

public abstract class AbstractObjDynamicHistogram<T>
extends ObjHistogram<T>

A dynamic histogram can dynamically adapt the number of bins to the data fed into the histogram.

Since:

0.5.5

Author:

Erich Schubert

Nested Class Summary

Nested classes/interfaces inherited from class elki.utilities.datastructures.histogram.ObjHistogram

ObjHistogram.BucketFactory<T>, ObjHistogram.Iter

Field Summary

Fields

Modifier and Type	Field	Description
private int	cachefill	Cache fill size
private double[]	cacheposs	Cache for positions to be inserted.
private java.lang.Object[]	cachevals	Cache for data to be inserted.
private int	destsize	Destination (minimum) size of the structure.

Fields inherited from class elki.utilities.datastructures.histogram.ObjHistogram

data, special, supplier

Fields inherited from class elki.utilities.datastructures.histogram.AbstractStaticHistogram

base, binsize, max, offset, size

Fields inherited from interface elki.utilities.datastructures.histogram.Histogram

CACHE_SHIFT

Constructor Summary

Constructors

Constructor	Description
AbstractObjDynamicHistogram(int bins)	Constructor.

Method Summary

Modifier and Type	Method	Description
protected abstract T	aggregate(T first, T second)	Rule to combine two bins or entries into one.
protected void	aggregateSpecial(T value, int bin)	Aggregate for a special value.
protected abstract T	cloneForCache(T data)	Clone a data passed to the algorithm for computing the initial size.
protected abstract T	downsample(java.lang.Object[] data, int start, int end, int size)	Perform downsampling on a number of bins.
T	get(double coord)	Access the value of a bin with new data.
double	getBinsize()	Get the size (width) of a bin.
double	getCoverMaximum()	Get maximum (covered by bins, not data!)
double	getCoverMinimum()	Get minimum (covered by bins, not data!)
int	getNumBins()	Get the number of bins actually in use.
ObjHistogram.Iter	iter()	Get an iterator over all histogram bins.
protected abstract T	makeObject()	Make a new empty bucket for the data store.
(package private) void	materialize()	Materialize the histogram from the cache.
void	putData(double coord, T value)	Put fresh data into the histogram (or into the cache).
private void	testResample(double coord)	Test (and perform) downsampling when needed.

Methods inherited from class elki.utilities.datastructures.histogram.ObjHistogram

getSpecial

Methods inherited from class elki.utilities.datastructures.histogram.AbstractStaticHistogram

getBinNr, growSize

Methods inherited from class java.lang.Object

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

Field Detail

cacheposs

private double[] cacheposs

Cache for positions to be inserted.

cachevals

private java.lang.Object[] cachevals

Cache for data to be inserted.

cachefill

private int cachefill

Cache fill size

destsize

private int destsize

Destination (minimum) size of the structure. At most 2*destsize bins are allowed.

Constructor Detail

AbstractObjDynamicHistogram

public AbstractObjDynamicHistogram(int bins)

Constructor.

Parameters:

bins - Design number of bins - may become twice as large!

Method Detail

materialize

void materialize()

Materialize the histogram from the cache.

get

public T get(double coord)

Description copied from class: ObjHistogram

Access the value of a bin with new data.

Overrides:

get in class ObjHistogram<T>

Parameters:

coord - Coordinate

Returns:

bin contents

putData

public void putData(double coord,
                    T value)

Put fresh data into the histogram (or into the cache).

Parameters:

coord - Coordinate

value - Value

aggregateSpecial

protected void aggregateSpecial(T value,
                                int bin)

Aggregate for a special value.

Parameters:

value - Parameter value

bin - Special bin index.

testResample

private void testResample(double coord)

Test (and perform) downsampling when needed.

Parameters:

coord - coordinate to accommodate.

iter

public ObjHistogram.Iter iter()

Description copied from class: AbstractStaticHistogram

Get an iterator over all histogram bins.

Specified by:

iter in interface Histogram

Overrides:

iter in class ObjHistogram<T>

Returns:

Iterator

getNumBins

public int getNumBins()

Description copied from class: AbstractStaticHistogram

Get the number of bins actually in use.

Specified by:

getNumBins in interface Histogram

Overrides:

getNumBins in class AbstractStaticHistogram

Returns:

number of bins

getBinsize

public double getBinsize()

Description copied from class: AbstractStaticHistogram

Get the size (width) of a bin.

Specified by:

getBinsize in interface Histogram

Overrides:

getBinsize in class AbstractStaticHistogram

Returns:

bin size

getCoverMinimum

public double getCoverMinimum()

Description copied from class: AbstractStaticHistogram

Get minimum (covered by bins, not data!)

Specified by:

getCoverMinimum in interface Histogram

Overrides:

getCoverMinimum in class AbstractStaticHistogram

Returns:

minimum

getCoverMaximum

public double getCoverMaximum()

Description copied from class: AbstractStaticHistogram

Get maximum (covered by bins, not data!)

Specified by:

getCoverMaximum in interface Histogram

Overrides:

getCoverMaximum in class AbstractStaticHistogram

Returns:

maximum

downsample

protected abstract T downsample(java.lang.Object[] data,
                                int start,
                                int end,
                                int size)

Perform downsampling on a number of bins.

Parameters:

data - Data array (needs cast!)

start - Interval start

end - Interval end (exclusive)

size - Intended size - extra bins are assumed to be empty, should be a power of two

Returns:

New bin value

aggregate

protected abstract T aggregate(T first,
                               T second)

Rule to combine two bins or entries into one.

Note: first and second MAY be modified and returned, they will not be used afterwards.

Parameters:

first - First bin value

second - Second bin value

Returns:

combined bin value

cloneForCache

protected abstract T cloneForCache(T data)

Clone a data passed to the algorithm for computing the initial size.

Parameters:

data - Data to be cloned

Returns:

cloned data

makeObject

protected abstract T makeObject()

Make a new empty bucket for the data store.

Returns:

New instance.