magpie.optimization.rankers

Class BaseEntryRanker

java.lang.Object

magpie.optimization.rankers.BaseEntryRanker

All Implemented Interfaces:

java.lang.Cloneable, Options

Direct Known Subclasses:

ClassProbabilityRanker, CompositionDistanceRanker, MultiObjectiveEntryRanker, PropertyRanker, SimpleEntryRanker, TargetEntryRanker

public abstract class BaseEntryRanker
extends java.lang.Object
implements java.lang.Cloneable, Options

This is a template for classes that rank entries based on some sort of objective function. When implementing this class, you only need to provide the objective function. The final user must decide whether it uses measured or predicted data and whether the goal is to maximize or minimize the objective function.

Implementation Guide

Only two operations and the Options interface need to be defined:

1. objectiveFunction(magpie.data.BaseEntry): Compute the value of an objective function based on an entry. This value is used to sort entries.
2. train(magpie.data.Dataset): Perform any operations required before running the objective function. Not always necessary.

Constructor Summary

Constructors

Constructor and Description
BaseEntryRanker()

Method Summary

Modifier and Type	Method and Description
BaseEntryRanker	clone()
boolean	isMaximizing()
boolean	isUsingMeasured()
abstract double	objectiveFunction(BaseEntry Entry) Some kind of objective function that returns a double when given an entry.
int[]	rankEntries(Dataset Data) Calculate the rank of entries based on this objective function
int[]	rankEntries(Dataset Data, boolean useMeasured) Calculate the rank of entries based on this objective function
int[]	rankEntries(Dataset Data, double[] Values) Return the rank of entries that were sorted based on the objective function defined in this class (also depends on the settings for Maximize and Measured.
double[]	runObjectiveFunction(Dataset Data) Evaluate every entry in a dataset using the objective function
void	setMaximizeFunction(boolean toMaximize) Set whether this entry ranker will maximize objective function
void	setUseMeasured(boolean useMeasured) Set whether to used measured (or predicted) class values
void	sortByRanking(Dataset data) Sort entries in a dataset according to ranking
abstract void	train(Dataset data) Train ranker.

Methods inherited from class java.lang.Object

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

Methods inherited from interface magpie.utility.interfaces.Options

printUsage, setOptions

Constructor Detail

BaseEntryRanker

public BaseEntryRanker()

Method Detail

clone

public BaseEntryRanker clone()

Overrides:

clone in class java.lang.Object

setUseMeasured

public void setUseMeasured(boolean useMeasured)

Set whether to used measured (or predicted) class values

Parameters:

useMeasured - Desired option

isUsingMeasured

public boolean isUsingMeasured()

Returns:

Whether this ranker is using measured or predicted class variables

setMaximizeFunction

public void setMaximizeFunction(boolean toMaximize)

Set whether this entry ranker will maximize objective function

Parameters:

toMaximize - Whether to maximize function

isMaximizing

public boolean isMaximizing()

Returns:

Whether the goal is to maximize the objective

objectiveFunction

public abstract double objectiveFunction(BaseEntry Entry)

Some kind of objective function that returns a double when given an entry.

NOTE: This function needs to support using either the measured or predicted class depending on the settings

Parameters:

Entry - Entry to be analyzed

Returns:

Evaluation of target function

rankEntries

public int[] rankEntries(Dataset Data,
                         double[] Values)

Return the rank of entries that were sorted based on the objective function defined in this class (also depends on the settings for Maximize and Measured. Will return the values for each instance (sorted in the same order as the output, which require preallocating Values.

double[] Values = new double[Data.NEntries()];

Parameters:

Data - Dataset containing entries to be ranked

Values - Pre-allocated, containing Data.NEntries() number of entries

Returns:

Entry number ranked by objective function

runObjectiveFunction

public double[] runObjectiveFunction(Dataset Data)

Evaluate every entry in a dataset using the objective function

Parameters:

Data - Dataset to be evaluated

Returns:

Value of objective function for each entry

rankEntries

public final int[] rankEntries(Dataset Data)

Calculate the rank of entries based on this objective function

Parameters:

Data - Dataset containing entries to be ranked

Returns:

Entry numbers descending in rank (0 is the best).

rankEntries

public final int[] rankEntries(Dataset Data,
                               boolean useMeasured)

Calculate the rank of entries based on this objective function

Parameters:

Data - Dataset containing entries to be ranked

useMeasured - Whether to use measured class variable

Returns:

Entry numbers descending in rank (0 is the best)

train

public abstract void train(Dataset data)

Train ranker.

Parameters:

data - Data to use as training set

sortByRanking

public void sortByRanking(Dataset data)

Sort entries in a dataset according to ranking

Parameters:

data - Dataset to be sorted