com.flaptor.hist4j

Class AdaptiveHistogram

Object

AdaptiveHistogram

All Implemented Interfaces:

Serializable

public class AdaptiveHistogram
extends Object
implements Serializable

This class implements a histogram that adapts to an unknown data distribution. It keeps a more or less constant resolution throughout the data range by increasing the resolution where the data is more dense. For example, if the data has such such a distribution that most of the values lie in the 0-5 range and only a few are in the 5-10 range, the histogram would adapt and assign more counting buckets to the 0-5 range and less to the 5-10 range. This implementation provides a method to obtain the accumulative density function for a given data point, and a method to obtain the data point that splits the data set at a given percentile.

Author:

Jorge Handl

See Also:

Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
protected static interface	AdaptiveHistogram.ValueConversion Auxiliary interface for inline functor object.

Constructor Summary

Constructors

Constructor and Description
AdaptiveHistogram() Class constructor.

Method Summary

Modifier and Type	Method and Description
void	addValue(double value) Adds a data point to the histogram.
long	getAccumCount(double value) Returns the cumulative density function for a given data point.
long	getCount(double value) Returns the number of data points stored in the same bucket as a given value.
protected int	getCountPerNodeLimit() This method is used by the internal data structure of the histogram to get the limit of data points that should be counted at one bucket.
double	getValueForPercentile(int percentile) Returns the data point that splits the data set at a given percentile.
void	normalize(double targetMin, double targetMax) Normalizes all the values to the desired range.
void	reset() Erases all data from the histogram.
void	show() Shows the histograms' underlying data structure.
ArrayList<Cell>	toTable() Return a table representing the data in this histogram.

Methods inherited from class Object

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

Constructor Detail

AdaptiveHistogram

public AdaptiveHistogram()

Class constructor.

Method Detail

reset

public void reset()

Erases all data from the histogram.

addValue

public void addValue(double value)

Adds a data point to the histogram.

Parameters:

value - the data point to add.

getCount

public long getCount(double value)

Returns the number of data points stored in the same bucket as a given value.

Parameters:

value - the reference data point.

Returns:

the number of data points stored in the same bucket as the reference point.

getAccumCount

public long getAccumCount(double value)

Returns the cumulative density function for a given data point.

Parameters:

value - the reference data point.

Returns:

the cumulative density function for the reference point.

getValueForPercentile

public double getValueForPercentile(int percentile)

Returns the data point that splits the data set at a given percentile.

Parameters:

percentile - the percentile at which the data set is split.

Returns:

the data point that splits the data set at the given percentile.

getCountPerNodeLimit

protected int getCountPerNodeLimit()

This method is used by the internal data structure of the histogram to get the limit of data points that should be counted at one bucket.

Returns:

the limit of data points to store a one bucket.

normalize

public void normalize(double targetMin,
                      double targetMax)

Normalizes all the values to the desired range.

Parameters:

targetMin - the target new minimum value.

targetMax - the target new maximum value.

show

public void show()

Shows the histograms' underlying data structure.

toTable

public ArrayList<Cell> toTable()

Return a table representing the data in this histogram. Each element is a table cell containing the range limit values and the count for that range.