weka.gui.beans

Class StreamThroughput

java.lang.Object

weka.gui.beans.StreamThroughput

All Implemented Interfaces:

java.io.Serializable

public class StreamThroughput
extends java.lang.Object
implements java.io.Serializable

Class for measuring throughput of an incremental Knowledge Flow step. Typical usage is to construct a StreamThroughput object at the start of the stream (i.e. FORMAT_AVAILABLE event) and then for each instance received call updateStart() just before processing the instance and then updateEnd() just after. If updateEnd() is called *before* sending any event to downstream step(s) then throughput just with respect to work done by the step will be measured. Elapsed time to process each instance (along with the number of instances) is accumulated over the sample time period. Instances per second is computed at the end of each sample period and added to a running total. Average instances/sec is reported to the status area of the log.

See Also:

Serialized Form

Field Summary

Fields

Modifier and Type	Field and Description
protected int	m_avInstsPerSec
protected double	m_cumulativeTime
protected int	m_instanceCount
protected int	m_numSamples
protected int	m_sampleCount
protected int	m_sampleTime sample period over which to count instances processed and instances/sec throughput
protected double	m_startTime
protected java.lang.String	m_statusMessagePrefix
protected double	m_updateStart

Constructor Summary

Constructors

Constructor and Description
StreamThroughput(java.lang.String statusMessagePrefix) Construct a new StreamThroughput
StreamThroughput(java.lang.String statusMessagePrefix, java.lang.String initialMessage, Logger log) Construct a new StreamThroughput

Method Summary

Modifier and Type	Method and Description
protected boolean	computeUpdate(double end)
java.lang.String	finished() Register the end of measurement.
java.lang.String	finished(Logger log) Register the end of measurement.
int	getAverageInstancesPerSecond() Get the average instances per second
void	setSamplePeriod(int period) Set the sampling period (in milliseconds) to compute througput over
void	updateEnd(Logger log) Register a throughput measurement end point.
void	updateStart() Register a throughput measurement start point

Methods inherited from class java.lang.Object

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

Field Detail

m_avInstsPerSec

protected transient int m_avInstsPerSec

m_startTime

protected transient double m_startTime

m_instanceCount

protected transient int m_instanceCount

m_sampleCount

protected transient int m_sampleCount

m_statusMessagePrefix

protected transient java.lang.String m_statusMessagePrefix

m_sampleTime

protected transient int m_sampleTime

sample period over which to count instances processed and instances/sec throughput

m_cumulativeTime

protected transient double m_cumulativeTime

m_numSamples

protected transient int m_numSamples

m_updateStart

protected transient double m_updateStart

Constructor Detail

StreamThroughput

public StreamThroughput(java.lang.String statusMessagePrefix)

Construct a new StreamThroughput

Parameters:

statusMessagePrefix - the unique identifier of the Knowledge Flow component being measured. This enables the correct line in the status area to be updated. See any Knowledge Flow step for an example.

StreamThroughput

public StreamThroughput(java.lang.String statusMessagePrefix,
                        java.lang.String initialMessage,
                        Logger log)

Construct a new StreamThroughput

Parameters:

statusMessagePrefix - the unique identifier of the Knowledge Flow component being measured. This enables the correct line in the status area to be updated. See any Knowledge Flow step for an example.

initialMessage - an initial message to print to the status area for this step on construction

log - the log to write status updates to

Method Detail

setSamplePeriod

public void setSamplePeriod(int period)

Set the sampling period (in milliseconds) to compute througput over

Parameters:

period - the sampling period in milliseconds

updateStart

public void updateStart()

Register a throughput measurement start point

updateEnd

public void updateEnd(Logger log)

Register a throughput measurement end point. Collects counts and statistics. Will update the status area for the KF step in question if the sample period has elapsed.

Parameters:

log - the log to write status updates to

computeUpdate

protected boolean computeUpdate(double end)

getAverageInstancesPerSecond

public int getAverageInstancesPerSecond()

Get the average instances per second

Returns:

the average instances per second processed

finished

public java.lang.String finished(Logger log)

Register the end of measurement. Writes a "Finished" update (that includes the final throughput info) to the status area of the log.

Parameters:

log - the log to write to

Returns:

the message written to the status area.

finished

public java.lang.String finished()

Register the end of measurement. Does not write a "Finished" update to the log

Returns:

a message that contains the final throughput info.