penalties.defined

Class PredictionConfidencePenalty

java.lang.Object

penalties.defined.PredictionConfidencePenalty

All Implemented Interfaces:

PenaltyFunction

public class PredictionConfidencePenalty
extends java.lang.Object
implements PenaltyFunction

PredictionConfidencePenalty penalty implements the cost function diverging from the original ranking. i.e. regularizing effect It is a ranking based penalty and does not support incremental updates. Please see the manuscript for the exact definition of this penalty.

Since:

0.0

Version:

1.0

Author:

Joeri Ruyssinck (joeri.ruysssinck@intec.ugent.be)

Constructor Summary

Constructors

Constructor and Description
PredictionConfidencePenalty(java.lang.Object argumentList) Constructs a new PredictionConfidencePenalty, the arguments are specified as an String[] passed as an object for classloading.

Method Summary

Modifier and Type	Method and Description
void	calculateMetricFromScratch() Calculates the metric from the current subnetwork/ranking
double	getCoef() Gets the relative weight for this penalty
double	getMetric() Returns the current penalty metric (unmapped to the cost)
java.lang.String	getPenaltyName() The penalty name, used for logging functionality
long	getPenaltyScore() Returns the current penalty score (metric mapped on cost function)
boolean	isGlobalPenalty() Indicates if this penalty works on the ranking or on (sub)network level
void	revertIncrementalPenalty() Reverts the penalty score to the previous value, discarding any changes made by the latest incremental update
void	setCoef(double coef) Sets the relative weight for this penalty
void	setNetwork(Network predictionNetwork) Sets the network/ranking associated with this penalty
void	updateIncrementalPenalty() Calculates the new penalty score in an incremental way after (small) changes have been to the network

Methods inherited from class java.lang.Object

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

Constructor Detail

PredictionConfidencePenalty

public PredictionConfidencePenalty(java.lang.Object argumentList)

Constructs a new PredictionConfidencePenalty, the arguments are specified as an String[] passed as an object for classloading.

Parameters:

argumentList - an Object in the form of a String array

Method Detail

getPenaltyName

public java.lang.String getPenaltyName()

Description copied from interface: PenaltyFunction

The penalty name, used for logging functionality

Specified by:

getPenaltyName in interface PenaltyFunction

Returns:

the penalty name

calculateMetricFromScratch

public void calculateMetricFromScratch()

Description copied from interface: PenaltyFunction

Calculates the metric from the current subnetwork/ranking

Specified by:

calculateMetricFromScratch in interface PenaltyFunction

getMetric

public double getMetric()

Description copied from interface: PenaltyFunction

Returns the current penalty metric (unmapped to the cost)

Specified by:

getMetric in interface PenaltyFunction

Returns:

the penalty metric

getPenaltyScore

public long getPenaltyScore()

Description copied from interface: PenaltyFunction

Returns the current penalty score (metric mapped on cost function)

Specified by:

getPenaltyScore in interface PenaltyFunction

Returns:

the penalty score

updateIncrementalPenalty

public void updateIncrementalPenalty()

Description copied from interface: PenaltyFunction

Calculates the new penalty score in an incremental way after (small) changes have been to the network

Specified by:

updateIncrementalPenalty in interface PenaltyFunction

revertIncrementalPenalty

public void revertIncrementalPenalty()

Description copied from interface: PenaltyFunction

Reverts the penalty score to the previous value, discarding any changes made by the latest incremental update

Specified by:

revertIncrementalPenalty in interface PenaltyFunction

isGlobalPenalty

public boolean isGlobalPenalty()

Description copied from interface: PenaltyFunction

Indicates if this penalty works on the ranking or on (sub)network level

Specified by:

isGlobalPenalty in interface PenaltyFunction

Returns:

true if the penalty works on the ranking

setNetwork

public void setNetwork(Network predictionNetwork)

Description copied from interface: PenaltyFunction

Sets the network/ranking associated with this penalty

Specified by:

setNetwork in interface PenaltyFunction

Parameters:

predictionNetwork - network or ranking associated with this penalty

getCoef

public double getCoef()

Description copied from interface: PenaltyFunction

Gets the relative weight for this penalty

Specified by:

getCoef in interface PenaltyFunction

Returns:

the relative weight

setCoef

public void setCoef(double coef)

Description copied from interface: PenaltyFunction

Sets the relative weight for this penalty

Specified by:

setCoef in interface PenaltyFunction

Parameters:

coef - the relative weight