meka.classifiers.multilabel.incremental

Class IncrementalEvaluation

java.lang.Object

meka.classifiers.multilabel.incremental.IncrementalEvaluation

public class IncrementalEvaluation
extends Object

IncrementalEvaluation.java - For Evaluating Incremental (Updateable) Classifiers.

Version:

September 2015

Author:

Jesse Read

Constructor Summary

Constructors

Constructor and Description
IncrementalEvaluation()

Method Summary

Modifier and Type	Method and Description
static Result	evaluateModel(MultiLabelClassifier h, weka.core.Instances D) EvaluateModel - over 20 windows.
static Result	evaluateModel(MultiLabelClassifier h, String[] options) EvaluateModel - Build and evaluate.
static Result	evaluateModelBatchWindow(MultiLabelClassifier h, weka.core.Instances D, int numWindows, double rLabeled, String Top, String Vop) EvaluateModelBatchWindow - Evaluate a multi-label data-stream model over windows.
static Result	evaluateModelPrequentialBasic(MultiLabelClassifier h, weka.core.Instances D, int windowSize, double rLabeled, String Top, String Vop) Prequential Evaluation - Accuracy since the start of evaluation.
static void	printOptions(Enumeration e)
static void	runExperiment(MultiLabelClassifier h, String[] args) RunExperiment - Build and evaluate a model with command-line options.

Methods inherited from class java.lang.Object

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

Constructor Detail

IncrementalEvaluation

public IncrementalEvaluation()

Method Detail

runExperiment

public static void runExperiment(MultiLabelClassifier h,
                                 String[] args)

RunExperiment - Build and evaluate a model with command-line options.

Parameters:

h - a multi-label updateable classifier

args - classifier + dataset options

evaluateModel

public static Result evaluateModel(MultiLabelClassifier h,
                                   String[] options)
                            throws Exception

EvaluateModel - Build and evaluate.

Parameters:

h - a multi-label Updateable classifier

options - dataset options (classifier options should already be set)

Returns:

The evaluation Result, null if none produced (eg when just outputting predictions)

Throws:

Exception

evaluateModel

public static Result evaluateModel(MultiLabelClassifier h,
                                   weka.core.Instances D)
                            throws Exception

EvaluateModel - over 20 windows.

Throws:

Exception

evaluateModelBatchWindow

public static Result evaluateModelBatchWindow(MultiLabelClassifier h,
                                              weka.core.Instances D,
                                              int numWindows,
                                              double rLabeled,
                                              String Top,
                                              String Vop)
                                       throws Exception

EvaluateModelBatchWindow - Evaluate a multi-label data-stream model over windows.

Parameters:

h - Multilabel Classifier

D - stream

numWindows - number of windows

rLabeled - labelled-ness (1.0 by default)

Top - threshold option

Vop - verbosity option

Returns:

The Result on the final window (but it contains samples of all the other evaluated windows). The window is sampled every N/numWindows instances, for a total of numWindows windows.

Throws:

Exception

evaluateModelPrequentialBasic

public static Result evaluateModelPrequentialBasic(MultiLabelClassifier h,
                                                   weka.core.Instances D,
                                                   int windowSize,
                                                   double rLabeled,
                                                   String Top,
                                                   String Vop)
                                            throws Exception

Prequential Evaluation - Accuracy since the start of evaluation.

Parameters:

h - Multilabel Classifier

D - stream

windowSize - sampling frequency (of evaluation statistics)

rLabeled - labelled-ness (1.0 by default)

Top - threshold option

Vop - verbosity option The window is sampled every N/numWindows instances, for a total of numWindows windows.

Throws:

Exception

printOptions

public static void printOptions(Enumeration e)