meka.core

Class MLEvalUtils

java.lang.Object

meka.core.MLEvalUtils

public abstract class MLEvalUtils
extends java.lang.Object

MLEvalUtils - Utility functions for Evaluation.

Version:

March 2014

Author:

Jesse Read

See Also:

Metrics

Constructor Summary

Constructors

Constructor and Description
MLEvalUtils()

Method Summary

Methods

Modifier and Type	Method and Description
static Result	averageResults(Result[] folds) AverageResults - Create a Result with the average of an array of Results by taking the average +/- standand deviation.
static java.util.HashMap<java.lang.String,java.lang.Double>	getMLStats(double[][] Rpred, int[][] Y, double[] t, java.lang.String vop) GetMLStats - Given predictions and corresponding true values and a threshold string, retreive statistics.
static java.util.HashMap<java.lang.String,java.lang.Double>	getMLStats(double[][] Rpred, int[][] Y, java.lang.String t, java.lang.String vop) GetMLStats - Given predictions and corresponding true values and a threshold string, retreive statistics.
static java.util.HashMap<java.lang.String,java.lang.Double>	getMTStats(double[][] Rpred, int[][] Y, java.lang.String vop) GetMTStats - Given multi-target predictions and corresponding true values, retreive evaluation statistics.
static java.lang.String	getThreshold(java.util.ArrayList<double[]> Y, weka.core.Instances D, java.lang.String top) GetThreshold - Get a threshold from a Threshold OPtion string 'top'.
static void	main(java.lang.String[] args) Main - can use this function for writing tests during development.

Methods inherited from class java.lang.Object

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

Constructor Detail

MLEvalUtils

public MLEvalUtils()

Method Detail

getThreshold

public static java.lang.String getThreshold(java.util.ArrayList<double[]> Y,
                            weka.core.Instances D,
                            java.lang.String top)
                                     throws java.lang.Exception

GetThreshold - Get a threshold from a Threshold OPtion string 'top'.

Parameters:

Y - label space; for calculating a threshold with PCut

D - training data; for calculating a threshold with PCut

top - Threshold OPtion (either "PCut1", "PCutL" or a real value e.g. "0.5" or L real values e.g. "[0.1, 0.2, 0.8]" for L = 3

Throws:

java.lang.Exception

getMLStats

public static java.util.HashMap<java.lang.String,java.lang.Double> getMLStats(double[][] Rpred,
                                                              int[][] Y,
                                                              java.lang.String t,
                                                              java.lang.String vop)

GetMLStats - Given predictions and corresponding true values and a threshold string, retreive statistics.

Parameters:

Rpred - predictions (may be real-valued confidences)

Y - corresponding true values

t - a threshold string, e.g. "0.387"

Returns:

the evaluation statistics

getMLStats

public static java.util.HashMap<java.lang.String,java.lang.Double> getMLStats(double[][] Rpred,
                                                              int[][] Y,
                                                              double[] t,
                                                              java.lang.String vop)

GetMLStats - Given predictions and corresponding true values and a threshold string, retreive statistics.

Parameters:

Rpred - predictions (may be double-valued confidences in the multi-label case)

Y - corresponding true values

t - a vector of thresholds, e.g. [0.1,0.1,0.1] or [0.1,0.5,0.4,0.001]

Returns:

the evaluation statistics

getMTStats

public static java.util.HashMap<java.lang.String,java.lang.Double> getMTStats(double[][] Rpred,
                                                              int[][] Y,
                                                              java.lang.String vop)

GetMTStats - Given multi-target predictions and corresponding true values, retreive evaluation statistics.

Parameters:

Rpred - predictions

Y - corresponding true values

Returns:

the evaluation statistics

averageResults

public static Result averageResults(Result[] folds)

AverageResults - Create a Result with the average of an array of Results by taking the average +/- standand deviation.

Parameters:

folds - array of Results (e.g., from CV-validation)

Returns:

A result reporting the average of these folds.

main

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

Main - can use this function for writing tests during development.

Parameters:

args - command line arguments