netkit.classifiers
Class DataSampler

java.lang.Object
  netkit.classifiers.DataSampler

public final class DataSampler
extends java.lang.Object

Field Summary

int

clsIdx

Constructor Summary

DataSampler(Node[] nodes, int attribIdx)

DataSampler(Node[] nodes, int attribIdx, long seed)

DataSampler(Node[] nodes, int attribIdx, long seed, boolean replacement, boolean stratified, boolean sampleUnknown)

Method Summary

static Graph	buildGraph()
DataSampler	clone()
Node[][][]	crossValidate(int numSplits) Create full cross-validation node sets in the form result[numsplit][0][...] is the training set for split numsplit and result[numsplit][1][...] is the test set for split numsplit.
boolean	doReplacement()
boolean	doStratified()
double[]	getDistribution()
static void	main(java.lang.String[] args)
int	numMissingValues()
Node[][]	sample(int... sizes) Sample sets of the given size from the underlying distribution of nodes and return them in a a list of lists.
Node[]	sample(int size) Sample the given number of nodes and return a new array filled with the samples.
void	sample(Node[]... result) Fill the given list of arrays with sample nodes, ensuring that there is no overlap between nodes sampled in each of the sub-arrays (unless you are sampling with replacement).
void	sample(Node[] result) Fill the array with sampled data
boolean	sampleUnknown()
int	size()
java.lang.String	toString()

Methods inherited from class java.lang.Object

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

Field Detail

clsIdx

public final int clsIdx

Constructor Detail

DataSampler

public DataSampler(Node[] nodes,
                   int attribIdx)

DataSampler

public DataSampler(Node[] nodes,
                   int attribIdx,
                   long seed)

DataSampler

public DataSampler(Node[] nodes,
                   int attribIdx,
                   long seed,
                   boolean replacement,
                   boolean stratified,
                   boolean sampleUnknown)

Method Detail

clone

public DataSampler clone()

Overrides:

clone in class java.lang.Object

numMissingValues

public int numMissingValues()

size

public int size()

getDistribution

public double[] getDistribution()

doReplacement

public boolean doReplacement()

sampleUnknown

public boolean sampleUnknown()

doStratified

public boolean doStratified()

sample

public void sample(Node[] result)

Fill the array with sampled data

Parameters:

result -

sample

public Node[] sample(int size)

Sample the given number of nodes and return a new array filled with the samples.

Parameters:

size -

Returns:

sample

public Node[][] sample(int... sizes)

Sample sets of the given size from the underlying distribution of nodes and return them in a a list of lists. This ensures that there is no overlap in nodes between the two arrays unless you are sampling with replacement).

NOTE: there is no guarantee of non-overlap if you are sampling with replacement.

Parameters:

sizes - The sizes of the lists to sample

Returns:

a list of lists of the given sizes.

sample

public void sample(Node[]... result)

Fill the given list of arrays with sample nodes, ensuring that there is no overlap between nodes sampled in each of the sub-arrays (unless you are sampling with replacement).

NOTE: there is no guarantee of non-overlap if you are sampling with replacement.

Parameters:

result -

crossValidate

public Node[][][] crossValidate(int numSplits)

Create full cross-validation node sets in the form result[numsplit][0][...] is the training set for split numsplit and result[numsplit][1][...] is the test set for split numsplit.

NOTE: there is no guarantee of non-overlap if you are sampling with replacement.

Parameters:

numSplits -

Returns:

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object

buildGraph

public static Graph buildGraph()

main

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