meka.core

Class SuperLabelUtils

java.lang.Object

meka.core.SuperLabelUtils

public abstract class SuperLabelUtils
extends java.lang.Object

SuperLabelUtils.java - Handy Utils for working with Meta Labels.
TODO call this class SuperClassUtils? SuperLabelUtils? Partition? PartitionUtils?

Version:

March 2014

Author:

Jesse Read

Constructor Summary

Constructors

Constructor and Description
SuperLabelUtils()

Method Summary

Methods

Modifier and Type	Method and Description
static int[][]	generatePartition(int L) generatePartition - return [[0],...,[L-1]].
static int[][]	generatePartition(int[] indices, int num, java.util.Random r) generatePartition.
static int[][]	generatePartition(int[] indices, int num, java.util.Random r, boolean balanced)
static int[][]	generatePartition(int[] indices, java.util.Random r) generatePartition - .
static int[]	get_k_subset(int L, int k, java.util.Random r) Get k subset - return a set of k label indices (of L possible labels).
static int[][]	getPartitionFromDatasetHierarchy(weka.core.Instances D) Get Partition From Dataset Hierarchy - assumes attributes are hierarchically arranged with '.'.
static weka.core.Instances	makePartitionDataset(weka.core.Instances D, int[] part) Make Partition Dataset - out of dataset D, on indices part[].
static weka.core.Instances	makePartitionDataset(weka.core.Instances D, int[] part, int P, int N) Make Partition Dataset - out of dataset D, on indices part[].
static java.lang.String	toString(int[][] partition) ToString - A string representation for the super-class partition 'partition'.

Methods inherited from class java.lang.Object

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

Constructor Detail

SuperLabelUtils

public SuperLabelUtils()

Method Detail

get_k_subset

public static int[] get_k_subset(int L,
                 int k,
                 java.util.Random r)

Get k subset - return a set of k label indices (of L possible labels).

generatePartition

public static int[][] generatePartition(int L)

generatePartition - return [[0],...,[L-1]].

Parameters:

L - number of labels

Returns:

[[0],...,[L-1]]

generatePartition

public static int[][] generatePartition(int[] indices,
                        java.util.Random r)

generatePartition - .

Parameters:

indices - [1,2,..,L]

r - Random

Returns:

partition

generatePartition

public static int[][] generatePartition(int[] indices,
                        int num,
                        java.util.Random r)

generatePartition.
TODO can generate 'indices' inside, given L
Get a random layout of 'num' sets of 'indices'.

Parameters:

indices - [1,2,...,L]

num - number of super-nodes to generate (between 1 and L)

r - Random

Returns:

partition

generatePartition

public static int[][] generatePartition(int[] indices,
                        int num,
                        java.util.Random r,
                        boolean balanced)

getPartitionFromDatasetHierarchy

public static final int[][] getPartitionFromDatasetHierarchy(weka.core.Instances D)

Get Partition From Dataset Hierarchy - assumes attributes are hierarchically arranged with '.'. For example europe.spain indicates leafnode spain of branch europe.

Parameters:

D - Dataset

Returns:

partition

toString

public static java.lang.String toString(int[][] partition)

ToString - A string representation for the super-class partition 'partition'.

makePartitionDataset

public static weka.core.Instances makePartitionDataset(weka.core.Instances D,
                                       int[] part)
                                                throws java.lang.Exception

Make Partition Dataset - out of dataset D, on indices part[].

Parameters:

D - regular multi-label dataset (of L = classIndex() labels)

part - list of indices we want to make into an LP dataset.

Returns:

Dataset with 1 multi-valued class label, representing the combinations of part[].

Throws:

java.lang.Exception

makePartitionDataset

public static weka.core.Instances makePartitionDataset(weka.core.Instances D,
                                       int[] part,
                                       int P,
                                       int N)
                                                throws java.lang.Exception

Make Partition Dataset - out of dataset D, on indices part[].

Parameters:

D - regular multi-label dataset (of L = classIndex() labels)

part - list of indices we want to make into a PS dataset.

P - see PSUtils

N - see PSUtils

Returns:

Dataset with 1 multi-valued class label, representing the combinations of part[].

Throws:

java.lang.Exception