ikr.simlib.statistics.median

Class MedianStatistic

java.lang.Object

ikr.simlib.statistics.Statistic

ikr.simlib.statistics.median.MedianStatistic

All Implemented Interfaces:

Printable

Direct Known Subclasses:

StdMedianStatistic

public abstract class MedianStatistic
extends Statistic
implements Printable

The classes MedianStatistic or StdMedianStatistic serve to register single measured values (e.g., run times) of the type double with the help of the method update(double).

The following functions are available for evaluation:

• Median value of all returned measured values (getMedian()) including a confidence interval (getMedianConfidenceInterval())
• Value range (getMinimum() and getMaximum()) in reference to all measured values
• Number of measured values from the batch (getBatchEvents())

If the parameter maxBufferSize is set to zero, all samples will be stored internally and the median will be calculated at the end of the simulation. Since this is very memory consuming, maxBufferSize can be set to a number > 0, which determines the maximum number of samples the statistic should hold in memory. The median is then calculated at the end of the simulation as an approximation that is based on all samples.

For printing the results it defines the following keywords:

median

Median value of all measured values in reference to all batches

cintm

confidence interval to median value

max

maximum sample of all batches

min

minimum sample of all batches

bmedian

number of samples in actual Batch

The calculation of the median usually needs to buffer all measured values. In order to save memory an approximation algorithm is served by the median statistic class:

• all values are saved in a list of value pairs of one double (value) and one integer value (weight)
• the weight is set to 1 for every new value added to the list
• when the list size exceeds the maximum buffer size, all entries are sorted by the value and grouped to pairs by adding their weight and calculating like this:
  value = value1 · weight1 + value2 · weight2
  
• When the method getMedian() is called the list is sorted again by value. Then the list is iterated from beginning until the sum of the weight-values is larger than the half of the total sum of weights. After this the found item and the previous item are selected and an linear approximation between the two value pairs is done.

See Also:

MedianTimeMeter

Field Summary

Fields inherited from class ikr.simlib.statistics.Statistic

sampleIndex, simNode, traceWriter

Constructor Summary

Constructors

Constructor and Description
MedianStatistic(SimNode ownNode)

Method Summary

Methods

Modifier and Type	Method and Description
abstract long	getBatchEvents()
abstract double	getBatchMedian()
abstract long	getEvents()
abstract double	getMaximum()
abstract double	getMedian()
abstract double	getMedianConfidenceInterval()
abstract double	getMinimum()
java.lang.String	getResultTypeName()
void	printResults(ResultType type, PrintResultWriter writer)
abstract void	update(double sample)

Methods inherited from class ikr.simlib.statistics.Statistic

addResultTag, addResultTag, computeMeasures, disableTracing, enableTracing, handleInitSimulation, handleStartBatch, handleStartTransientPhase, handleStopBatch, handleStopTransientPhase, printComments, resetBatchStatistic, resetStatistic, update, writeTraceEntry

Methods inherited from class java.lang.Object

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

Constructor Detail

MedianStatistic

public MedianStatistic(SimNode ownNode)

Method Detail

printResults

public void printResults(ResultType type,
                PrintResultWriter writer)

Specified by:

printResults in interface Printable

getResultTypeName

public java.lang.String getResultTypeName()

Specified by:

getResultTypeName in interface Printable

update

public abstract void update(double sample)

getMedian

public abstract double getMedian()

getMedianConfidenceInterval

public abstract double getMedianConfidenceInterval()

getEvents

public abstract long getEvents()

getMaximum

public abstract double getMaximum()

getMinimum

public abstract double getMinimum()

getBatchMedian

public abstract double getBatchMedian()

getBatchEvents

public abstract long getBatchEvents()