weka.distributed

Class WekaClassifierEvaluationMapTask

java.lang.Object

weka.distributed.WekaClassifierEvaluationMapTask

All Implemented Interfaces:

java.io.Serializable

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

Map task for evaluating a trained classifier. Uses Instances.train/testCV() methods for extracting folds for batch trained classifiers or modulus operation for incrementally trained classifiers.

Version:

$Revision: 12342 $

Author:

Mark Hall (mhall{[at]}pentaho{[dot]}com)

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
WekaClassifierEvaluationMapTask()

Method Summary

Modifier and Type	Method and Description
void	finalizeTask() Finalize this task.
boolean	getBatchTrainedIncremental() Get whether the classifier is an incremental one that has been batch trained
weka.classifiers.Classifier	getClassifier() Get the classifier used
weka.classifiers.evaluation.Evaluation	getEvaluation() Get the evaluation object
int	getFoldNumber() Get the fold number to evaluate on
int	getTotalNumFolds() Get the total number of folds
static void	main(java.lang.String[] args)
void	processInstance(weka.core.Instance inst) Process an instance for evaluation
void	setBatchTrainedIncremental(boolean b) Set whether the classifier is an incremental one that has been batch trained
void	setClassifier(weka.classifiers.Classifier c) Set the classifier to use
void	setFoldNumber(int fn) Set the fold number to evaluate on.
void	setTotalNumFolds(int t) Set the total number of folds (1 for evaluating on all the data)
void	setup(weka.core.Instances trainingHeader, double[] priors, double count, long seed, double predFrac) Setup the task.

Methods inherited from class java.lang.Object

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

Constructor Detail

WekaClassifierEvaluationMapTask

public WekaClassifierEvaluationMapTask()

Method Detail

getEvaluation

public weka.classifiers.evaluation.Evaluation getEvaluation()

Get the evaluation object

Returns:

the evaluation object

setClassifier

public void setClassifier(weka.classifiers.Classifier c)

Set the classifier to use

Parameters:

c - the classifier to use

getClassifier

public weka.classifiers.Classifier getClassifier()

Get the classifier used

Returns:

the classifier used

setFoldNumber

public void setFoldNumber(int fn)

Set the fold number to evaluate on. -1 indicates that all the data should be used for evaluation

Parameters:

fn - the fold number to evaluate on or -1 to evaluate on all the data

getFoldNumber

public int getFoldNumber()

Get the fold number to evaluate on

Returns:

the fold number to evaluate on

setTotalNumFolds

public void setTotalNumFolds(int t)

Set the total number of folds (1 for evaluating on all the data)

Parameters:

t - the total number of folds

getTotalNumFolds

public int getTotalNumFolds()

Get the total number of folds

Returns:

the total number of folds

setBatchTrainedIncremental

public void setBatchTrainedIncremental(boolean b)

Set whether the classifier is an incremental one that has been batch trained

Parameters:

b - true if the classifier is incremental and it has been batch trained.

getBatchTrainedIncremental

public boolean getBatchTrainedIncremental()

Get whether the classifier is an incremental one that has been batch trained

Returns:

true if the classifier is incremental and it has been batch trained.

setup

public void setup(weka.core.Instances trainingHeader,
                  double[] priors,
                  double count,
                  long seed,
                  double predFrac)
           throws java.lang.Exception

Setup the task.

Parameters:

trainingHeader - the header of the training data used to create the classifier to be evaluated

priors - priors for the class (frequency counts for the values of a nominal class or sum of target for a numeric class)

count - the total number of class values seen (with respect to the priors)

seed - the random seed to use for shuffling the data in the batch case

predFrac - the fraction of the total number of predictions to retain for computing AUC/AUPRC

Throws:

java.lang.Exception - if a problem occurs

processInstance

public void processInstance(weka.core.Instance inst)
                     throws java.lang.Exception

Process an instance for evaluation

Parameters:

inst - the instance to process

Throws:

java.lang.Exception - if a problem occurs

finalizeTask

public void finalizeTask()
                  throws java.lang.Exception

Finalize this task. This is where the actual evaluation occurs in the batch case - the order of the data gets randomized (and stratified if class is nominal) and the test fold extracted.

Throws:

java.lang.Exception - if a problem occurs

main

public static void main(java.lang.String[] args)