de.jstacs.data
Class Sample

java.lang.Object
  de.jstacs.data.Sample

public class Sample
extends Object

This is the class for any sample of sequences. All sequences in a sample have to have the same AlphabetContainer. The sequences may have different lengths.
For the internal representation the class Sequence is used, where the external alphabet is converted to integral numerical values. The class Sample knows about this coding via instances of class AlphabetContainer respectively Alphabet.

There are different ways to access the elements of a Sample. If one needs random access there's the method getElementAt( int i ). For fast sequential access it is recommended to use an ElementEnumerator.

Sample is immutable.

Author:

Jens Keilwagen

See Also:

AlphabetContainer, Sequence

Nested Class Summary

static class	Sample.ElementEnumerator This class can be used to have a fast sequential access to a sample.
static class	Sample.PartitionMethod This enum defines different partition methods for a sample.
static class	Sample.WeightedSampleFactory This class enables you to eliminate sequences that occur more than once in one or more samples.

Constructor Summary

Sample(AlphabetContainer abc, AbstractStringExtractor se)
Creates a Sample from a StringExtractor using the given AlphabetContainer.

Sample(AlphabetContainer abc, AbstractStringExtractor se, int subsequenceLength)
Creates a Sample from a StringExtractor using the given AlphabetContainer and all overlapping windows of subsequenceLength.

Sample(AlphabetContainer abc, AbstractStringExtractor se, String delim)
Creates a Sample from a StringExtractor using the given AlphabetContainer and delimiter.

Sample(AlphabetContainer abc, AbstractStringExtractor se, String delim, int subsequenceLength)
Creates a Sample from a StringExtractor using the given AlphabetContainer, the given delimiter and all overlapping windows of subsequenceLength.

Sample(Sample s, int subsequenceLength)
This constructor enables you to use subsequences of the elements of a sample.

Sample(String annotation, Sequence... seqs)
This constructor is specially designed for the method Model.emitSample(int, int...).

Method Summary

Sequence[]	getAllElements() Returns an array of sequences containing all elements of this Sample.
AlphabetContainer	getAlphabetContainer() Returns the AlphabetContainer of this Sample.
String	getAnnotation() This method returns some annotation of the Sample.
static String	getAnnotation(Sample... s) Returns the annotation for an array of Samples.
Sample	getCompositeSample(int[] starts, int[] lengths) This method enables you to use only composite sequences of all elements in the current sample.
Sequence	getElementAt(int i) This method returns the element with index i.
int	getElementLength() Returns the length of the elements in this Sample.
Sample	getInfixSample(int start, int length) This method enables you to use only an infix of all elements in the current sample.
int	getMaximalElementLength() Returns the maximal length of an element in this Sample.
int	getMinimalElementLength() Returns the minimal length of an element in this Sample.
int	getNumberOfElements() Returns the number of elements in this Sample.
int	getNumberOfElementsWithLength(int len) Returns the number of overlapping elements that can be extracted.
Sample	getSuffixSample(int start) This method enables you to use only a suffix of all elements in the current sample.
static Sample	intersection(Sample... samples) This method computes the intersection between all elements of the array, i.e. it returns a Sample containing only sequences that are contained in all Samples of the array.
boolean	isDiscreteSample() This method returns true if all positions use discrete values.
boolean	isSimpleSample() This method answers the question whether all random variables are defined over the same range, i.e. all positions use the same (fixed) alphabet.
Sample[]	partition(double p, Sample.PartitionMethod method, int subsequenceLength) This method partitions the elements of the sample in 2 distinct parts.
Sample[]	partition(int k, Sample.PartitionMethod method) This method partitions the elements of the sample in k distinct parts.
Sample[]	partition(Sample.PartitionMethod method, double... percentage) This method partitions the elements of the sample in distinct parts.
void	save(String msg, File f) This method writes a message msg and the sample to a file f.
Sample	subSampling(int number) Randomly samples elements (sequences) from the set of all elements (sequences) contained in this Sample.
String	toString()
static Sample	union(Sample... s) Unites all samples in s.
static Sample	union(Sample[] s, boolean[] in) This method unites all Samples from s regarding in.
static Sample	union(Sample[] s, boolean[] in, int subsequenceLength) This method unites all Samples from s regarding in and sets the element length in the united sample to subsequenceLength.
static Sample	union(Sample[] s, int subsequenceLength) This method unites all Samples from s and sets the element length in the united sample to subsequenceLength.

Methods inherited from class java.lang.Object

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

Constructor Detail

Sample

public Sample(AlphabetContainer abc,
              AbstractStringExtractor se)
       throws WrongAlphabetException,
              EmptySampleException,
              WrongLengthException

Creates a Sample from a StringExtractor using the given AlphabetContainer.

Parameters:

abc - the AlphabetContainer

se - the StringExtractor

Throws:

WrongAlphabetException - if the AlphabetContainer is not suitable

EmptySampleException - if the Sample would be empty

WrongLengthException - never happens

Sample

public Sample(AlphabetContainer abc,
              AbstractStringExtractor se,
              int subsequenceLength)
       throws WrongAlphabetException,
              WrongLengthException,
              EmptySampleException

Creates a Sample from a StringExtractor using the given AlphabetContainer and all overlapping windows of subsequenceLength.

Parameters:

abc - the AlphabetContainer

se - the StringExtractor

subsequenceLength - the length of the window, sliding on the String of se. If len is 0 (zero) than the sequences are used as given from the StringExtractor

Throws:

WrongAlphabetException - if the AlphabetContainer is not suitable

WrongLengthException - if the subsequence length is not supported

EmptySampleException - if the Sample would be empty

Sample

public Sample(AlphabetContainer abc,
              AbstractStringExtractor se,
              String delim)
       throws WrongAlphabetException,
              EmptySampleException,
              WrongLengthException

Creates a Sample from a StringExtractor using the given AlphabetContainer and delimiter.

Parameters:

abc - the AlphabetContainer

se - the StringExtractor

delim - the delimiter for parsing the Strings

Throws:

WrongAlphabetException - if the AlphabetContainer is not suitable

EmptySampleException - if the Sample would be empty

WrongLengthException - never happens

Sample

public Sample(AlphabetContainer abc,
              AbstractStringExtractor se,
              String delim,
              int subsequenceLength)
       throws EmptySampleException,
              WrongAlphabetException,
              WrongLengthException

Creates a Sample from a StringExtractor using the given AlphabetContainer, the given delimiter and all overlapping windows of subsequenceLength.

Parameters:

abc - the AlphabetContainer

se - the StringExtractor

delim - the delimiter for parsing the Strings

subsequenceLength - the length of the window, sliding on the String of se. If len is 0 (zero) than the sequences are used as given from the StringExtractor

Throws:

WrongAlphabetException - if the AlphabetContainer is not suitable

EmptySampleException - if the Sample would be empty

WrongLengthException - if the subsequence length is not supported

Sample

public Sample(Sample s,
              int subsequenceLength)
       throws WrongLengthException

This constructor enables you to use subsequences of the elements of a sample.

It can also be used to ensure that all sequences that can be accessed by getElementAt(int) are real objects and do not have to be created at the invocation of the method. (The same holds for the Sample.ElementEnumerator. In those cases both ways to access the sequence are approximately equally fast.)

Parameters:

s - the sample

subsequenceLength - the new element length

Throws:

WrongLengthException - if something is wrong with subsequenceLength

Sample

public Sample(String annotation,
              Sequence... seqs)
       throws EmptySampleException,
              IllegalArgumentException

This constructor is specially designed for the method Model.emitSample(int, int...).

Parameters:

annotation - the annotation of the sample

seqs - the sequence(s)

Throws:

EmptySampleException - if the array seqs is null or the length is 0

IllegalArgumentException - if the alphabets do not match

Method Detail

getAnnotation

public static final String getAnnotation(Sample... s)

Returns the annotation for an array of Samples.

Parameters:

s - an array of Samples

Returns:

the annotation

intersection

public static final Sample intersection(Sample... samples)
                                 throws IllegalArgumentException,
                                        EmptySampleException

This method computes the intersection between all elements of the array, i.e. it returns a Sample containing only sequences that are contained in all Samples of the array.

Parameters:

samples - the array

Returns:

the intersection of the elements in the array

Throws:

IllegalArgumentException - if the elements of the array are from different domains

EmptySampleException - if the intersection is empty

union

public static final Sample union(Sample[] s,
                                 boolean[] in)
                          throws IllegalArgumentException,
                                 EmptySampleException

This method unites all Samples from s regarding in.

Parameters:

s - the Samples

in - an array indicating which sample is used in the union, if in[i]==true the sample s[i] is used

Returns:

the united sample

Throws:

IllegalArgumentException - if s.length != in.length or the alphabets do not match

EmptySampleException - if the union is empty

union

public static final Sample union(Sample... s)
                          throws IllegalArgumentException

Unites all samples in s.

Parameters:

s - the samples

Returns:

the united sample

Throws:

IllegalArgumentException - if the alphabets do not match

See Also:

union(Sample[], boolean[])

union

public static final Sample union(Sample[] s,
                                 boolean[] in,
                                 int subsequenceLength)
                          throws IllegalArgumentException,
                                 EmptySampleException,
                                 WrongLengthException

This method unites all Samples from s regarding in and sets the element length in the united sample to subsequenceLength.

Parameters:

s - the Samples

in - an array indicating which sample is used in the union, if in[i]==true the sample s[i] is used

subsequenceLength - the length of the elements in the united sample

Returns:

the united sample

Throws:

IllegalArgumentException - if s.length != in.length or the alphabets do not match

EmptySampleException - if the union is empty

WrongLengthException - if the united sample does not support this subsequenceLength

union

public static final Sample union(Sample[] s,
                                 int subsequenceLength)
                          throws IllegalArgumentException,
                                 WrongLengthException

This method unites all Samples from s and sets the element length in the united sample to subsequenceLength.

Parameters:

s - the Samples

subsequenceLength - the length of the elements in the united sample

Returns:

the united sample

Throws:

IllegalArgumentException - if the alphabets do not match

WrongLengthException - if the united sample does not support this subsequenceLength

See Also:

union(Sample[], boolean[], int)

getAllElements

public Sequence[] getAllElements()

Returns an array of sequences containing all elements of this Sample.

Returns:

all elements (sequences) of this Sample

getAlphabetContainer

public final AlphabetContainer getAlphabetContainer()

Returns the AlphabetContainer of this Sample.

Returns:

the AlphabetContainer of this Sample

getAnnotation

public final String getAnnotation()

This method returns some annotation of the Sample.

Returns:

some annotation of the Sample

getCompositeSample

public final Sample getCompositeSample(int[] starts,
                                       int[] lengths)
                                throws IllegalArgumentException

This method enables you to use only composite sequences of all elements in the current sample. The composite sequences will be returned in an new sample.

Parameters:

starts - the start positions

lengths - the lengths of the chunks

Returns:

a composite sample

Throws:

IllegalArgumentException - if either starts or lengths or both in combination are not suitable

getElementAt

public Sequence getElementAt(int i)

This method returns the element with index i. See also this comment.

Parameters:

i - the index

Returns:

the element

getElementLength

public int getElementLength()

Returns the length of the elements in this Sample.

Returns:

the length of the elements in this Sample

getInfixSample

public final Sample getInfixSample(int start,
                                   int length)
                            throws IllegalArgumentException

This method enables you to use only an infix of all elements in the current sample. The subsequences will be returned in an new sample.

This method can also be used to create a sample of prefixes if the element length is not zero.

Parameters:

start - the start position of the infix

length - the length of the infix, has to be positive

Returns:

a sample of specified infixes

Throws:

IllegalArgumentException - if either start or length or both in combination are not suitable

getMinimalElementLength

public int getMinimalElementLength()

Returns the minimal length of an element in this Sample.

Returns:

the minimal length of an element in this Sample

getMaximalElementLength

public int getMaximalElementLength()

Returns the maximal length of an element in this Sample.

Returns:

the maximal length of an element in this Sample

getNumberOfElements

public int getNumberOfElements()

Returns the number of elements in this Sample.

Returns:

the number of elements in this Sample

getNumberOfElementsWithLength

public int getNumberOfElementsWithLength(int len)
                                  throws WrongLengthException

Returns the number of overlapping elements that can be extracted.

Parameters:

len - the length of the elements

Returns:

the number of elements with the specified length

Throws:

WrongLengthException - if the given length is bigger than the minimal element length

getSuffixSample

public final Sample getSuffixSample(int start)
                             throws IllegalArgumentException

This method enables you to use only a suffix of all elements in the current sample. The subsequences will be returned in an new sample.

Parameters:

start - the start position of the suffix

Returns:

a sample of specified suffixes

Throws:

IllegalArgumentException - if start is not suitable

isSimpleSample

public final boolean isSimpleSample()

This method answers the question whether all random variables are defined over the same range, i.e. all positions use the same (fixed) alphabet.

Returns:

true if the sample is simple

isDiscreteSample

public final boolean isDiscreteSample()

This method returns true if all positions use discrete values.

Returns:

true if the sample is discrete

partition

public Sample[] partition(double p,
                          Sample.PartitionMethod method,
                          int subsequenceLength)
                   throws WrongLengthException,
                          UnsupportedOperationException,
                          EmptySampleException

This method partitions the elements of the sample in 2 distinct parts. The second part (test sample) holds the percentage of p, the first the rest (train sample). The first part has element length as the current sample, the second has element length subsequenceLength.

Parameters:

p - the percentage for the second part, the second part holds at least this percentage of the full sample

method - the method how to partition the sample (partitioning criterion)

subsequenceLength - the element length of the second part. If len is 0 (zero) than the sequences are used as given in this Sample.

Returns:

the partitioned sample

Throws:

WrongLengthException - if some is wrong with subsequenceLength

UnsupportedOperationException - if the sample is not simple

EmptySampleException - if at least one of the created partitions is empty

See Also:

Sample.PartitionMethod.PARTITION_BY_NUMBER_OF_ELEMENTS, Sample.PartitionMethod.PARTITION_BY_NUMBER_OF_SYMBOLS

partition

public Sample[] partition(Sample.PartitionMethod method,
                          double... percentage)
                   throws IllegalArgumentException,
                          EmptySampleException

This method partitions the elements of the sample in distinct parts.

Parameters:

method - the method how to partition the sample (partitioning criterion)

percentage - the array of percentages for each "subsample"

Returns:

the array of "subsamples"

Throws:

IllegalArgumentException - if something with the percentages is not correct (sum != 1 or one value is not in [0,1])

EmptySampleException - if at least one of the created partitions is empty

See Also:

Sample.PartitionMethod.PARTITION_BY_NUMBER_OF_ELEMENTS, Sample.PartitionMethod.PARTITION_BY_NUMBER_OF_SYMBOLS

partition

public Sample[] partition(int k,
                          Sample.PartitionMethod method)
                   throws IllegalArgumentException,
                          EmptySampleException

This method partitions the elements of the sample in k distinct parts.

Parameters:

k - the number of parts

method - how to split the data

Returns:

the array of "subsamples"

Throws:

IllegalArgumentException - if k is not correct

EmptySampleException - if at least one of the created partitions is empty

See Also:

Sample.PartitionMethod.PARTITION_BY_NUMBER_OF_ELEMENTS, Sample.PartitionMethod.PARTITION_BY_NUMBER_OF_SYMBOLS

subSampling

public Sample subSampling(int number)
                   throws EmptySampleException

Randomly samples elements (sequences) from the set of all elements (sequences) contained in this Sample.
Depending on whether this Sample is chosen to contain overlapping elements (windows of length subsequenceLength) or not, those elements (overlapping windows, whole sequences) are subsampled.

Parameters:

number - of Sequences that should be drawn from the contained set of sequences (with replacement)

Returns:

a new Sample containing the drawn Sequences

Throws:

EmptySampleException - if number is not positive

save

public final void save(String msg,
                       File f)
                throws IOException

This method writes a message msg and the sample to a file f.

Parameters:

msg - the message, any information

f - the File

Throws:

IOException - if something went wrong with the file

toString

public String toString()

Overrides:

toString in class Object