weka.filters
Class Filter

java.lang.Object
  weka.filters.Filter

All Implemented Interfaces:

java.io.Serializable

Direct Known Subclasses:

AbstractTimeSeries, Add, AddCluster, AddExpression, AddNoise, AllFilter, AttributeSelection, ClassOrder, ClusterMembership, Copy, Discretize, FirstOrder, MakeIndicator, MergeTwoValues, NominalToBinary, NominalToBinary, NonSparseToSparse, NullFilter, NumericTransform, Obfuscate, PotentialClassIgnorer, Randomize, RandomProjection, Remove, RemoveFolds, RemoveMisclassified, RemovePercentage, RemoveRange, RemoveType, RemoveUseless, RemoveWithValues, Resample, Resample, SparseToNonSparse, SpreadSubsample, StratifiedRemoveFolds, StringToNominal, StringToWordVector, SwapValues

public abstract class Filter

extends java.lang.Object

implements java.io.Serializable

An abstract class for instance filters: objects that take instances as input, carry out some transformation on the instance and then output the instance. The method implementations in this class assume that most of the work will be done in the methods overridden by subclasses.

A simple example of filter use. This example doesn't remove instances from the output queue until all instances have been input, so has higher memory consumption than an approach that uses output instances as they are made available:

  Filter filter = ..some type of filter..
  Instances instances = ..some instances..
  for (int i = 0; i < data.numInstances(); i++) {
    filter.input(data.instance(i));
  }
  filter.batchFinished();
  Instances newData = filter.outputFormat();
  Instance processed;
  while ((processed = filter.output()) != null) {
    newData.add(processed);
  }
  ..do something with newData..
 

Version:

$Revision: 1.23 $

Author:

Len Trigg (trigg@cs.waikato.ac.nz)

See Also:

Serialized Form

Field Summary

private boolean	m_Debug Debugging mode
private Instances	m_InputFormat The input format for instances
private int[]	m_InputStringAtts Indices of string attributes in the input format
protected boolean	m_NewBatch Record whether the filter is at the start of a batch
private Instances	m_OutputFormat The output format for instances
private Queue	m_OutputQueue The output instance queue
private int[]	m_OutputStringAtts Indices of string attributes in the output format

Constructor Summary

Filter()

Method Summary

static void	batchFilterFile(Filter filter, java.lang.String[] options) Method for testing filters ability to process multiple batches.
boolean	batchFinished() Signify that this batch of input to the filter is finished.
protected void	bufferInput(Instance instance) Adds the supplied input instance to the inputformat dataset for later processing.
protected void	copyStringValues(Instance instance, boolean instSrcCompat, Instances srcDataset, Instances destDataset) Takes string values referenced by an Instance and copies them from a source dataset to a destination dataset.
protected void	copyStringValues(Instance instance, boolean instSrcCompat, Instances srcDataset, int[] srcStrAtts, Instances destDataset, int[] destStrAtts) Takes string values referenced by an Instance and copies them from a source dataset to a destination dataset.
private void	copyStringValues(Instance inst, Instances destDataset, int[] strAtts) Copies string values contained in the instance copied to a new dataset.
static void	filterFile(Filter filter, java.lang.String[] options) Method for testing filters.
protected void	flushInput() This will remove all buffered instances from the inputformat dataset.
protected Instances	getInputFormat() Gets the currently set inputformat instances.
protected int[]	getInputStringIndex() Returns an array containing the indices of all string attributes in the input format.
Instances	getOutputFormat() Gets the format of the output instances.
protected int[]	getOutputStringIndex() Returns an array containing the indices of all string attributes in the output format.
protected int[]	getStringIndices(Instances insts) Gets an array containing the indices of all string attributes.
boolean	input(Instance instance) Input an instance for filtering.
boolean	inputFormat(Instances instanceInfo) Deprecated. use setInputFormat(Instances) instead.
protected Instances	inputFormatPeek() Returns a reference to the current input format without copying it.
boolean	isOutputFormatDefined() Returns whether the output format is ready to be collected
static void	main(java.lang.String[] args) Main method for testing this class.
int	numPendingOutput() Returns the number of instances pending output
Instance	output() Output an instance after filtering and remove from the output queue.
Instances	outputFormat() Deprecated. use getOutputFormat() instead.
protected Instances	outputFormatPeek() Returns a reference to the current output format without copying it.
Instance	outputPeek() Output an instance after filtering but do not remove from the output queue.
protected void	push(Instance instance) Adds an output instance to the queue.
protected void	resetQueue() Clears the output queue.
boolean	setInputFormat(Instances instanceInfo) Sets the format of the input instances.
protected void	setOutputFormat(Instances outputFormat) Sets the format of output instances.
static Instances	useFilter(Instances data, Filter filter) Filters an entire set of instances through a filter and returns the new set.

Methods inherited from class java.lang.Object

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

Field Detail

m_Debug

private boolean m_Debug

Debugging mode

m_OutputFormat

private Instances m_OutputFormat

The output format for instances

m_OutputQueue

private Queue m_OutputQueue

The output instance queue

m_OutputStringAtts

private int[] m_OutputStringAtts

Indices of string attributes in the output format

m_InputStringAtts

private int[] m_InputStringAtts

Indices of string attributes in the input format

m_InputFormat

private Instances m_InputFormat

The input format for instances

m_NewBatch

protected boolean m_NewBatch

Record whether the filter is at the start of a batch

Constructor Detail

Filter

public Filter()

Method Detail

setOutputFormat

protected void setOutputFormat(Instances outputFormat)

Sets the format of output instances. The derived class should use this method once it has determined the outputformat. The output queue is cleared.

Parameters:

outputFormat - the new output format

getInputFormat

protected Instances getInputFormat()

Gets the currently set inputformat instances. This dataset may contain buffered instances.

Returns:

the input Instances.

inputFormatPeek

protected Instances inputFormatPeek()

Returns a reference to the current input format without copying it.

Returns:

a reference to the current input format

outputFormatPeek

protected Instances outputFormatPeek()

Returns a reference to the current output format without copying it.

Returns:

a reference to the current output format

push

protected void push(Instance instance)

Adds an output instance to the queue. The derived class should use this method for each output instance it makes available.

Parameters:

instance - the instance to be added to the queue.

resetQueue

protected void resetQueue()

Clears the output queue.

bufferInput

protected void bufferInput(Instance instance)

Adds the supplied input instance to the inputformat dataset for later processing. Use this method rather than getInputFormat().add(instance). Or else. Note that the provided instance gets copied when buffered.

Parameters:

instance - the Instance to buffer.

getInputStringIndex

protected int[] getInputStringIndex()

Returns an array containing the indices of all string attributes in the input format. This index is created during setInputFormat()

Returns:

an array containing the indices of string attributes in the input dataset.

getOutputStringIndex

protected int[] getOutputStringIndex()

Returns an array containing the indices of all string attributes in the output format. This index is created during setOutputFormat()

Returns:

an array containing the indices of string attributes in the output dataset.

copyStringValues

private void copyStringValues(Instance inst,
                              Instances destDataset,
                              int[] strAtts)

Copies string values contained in the instance copied to a new dataset. The Instance must already be assigned to a dataset. This dataset and the destination dataset must have the same structure.

Parameters:

destDataset - the destination set of Instances

strAtts - an array containing the indices of any string attributes in the dataset.

copyStringValues

protected void copyStringValues(Instance instance,
                                boolean instSrcCompat,
                                Instances srcDataset,
                                Instances destDataset)

Takes string values referenced by an Instance and copies them from a source dataset to a destination dataset. The instance references are updated to be valid for the destination dataset. The instance may have the structure (i.e. number and attribute position) of either dataset (this affects where references are obtained from). The source dataset must have the same structure as the filter input format and the destination must have the same structure as the filter output format.

Parameters:

instance - the instance containing references to strings in the source dataset that will have references updated to be valid for the destination dataset.

instSrcCompat - true if the instance structure is the same as the source, or false if it is the same as the destination

srcDataset - the dataset for which the current instance string references are valid (after any position mapping if needed)

destDataset - the dataset for which the current instance string references need to be inserted (after any position mapping if needed)

copyStringValues

protected void copyStringValues(Instance instance,
                                boolean instSrcCompat,
                                Instances srcDataset,
                                int[] srcStrAtts,
                                Instances destDataset,
                                int[] destStrAtts)

Takes string values referenced by an Instance and copies them from a source dataset to a destination dataset. The instance references are updated to be valid for the destination dataset. The instance may have the structure (i.e. number and attribute position) of either dataset (this affects where references are obtained from). Only works if the number of string attributes is the same in both indices (implicitly these string attributes should be semantically same but just with shifted positions).

Parameters:

instance - the instance containing references to strings in the source dataset that will have references updated to be valid for the destination dataset.

instSrcCompat - true if the instance structure is the same as the source, or false if it is the same as the destination (i.e. which of the string attribute indices contains the correct locations for this instance).

srcDataset - the dataset for which the current instance string references are valid (after any position mapping if needed)

srcStrAtts - an array containing the indices of string attributes in the source datset.

destDataset - the dataset for which the current instance string references need to be inserted (after any position mapping if needed)

destStrAtts - an array containing the indices of string attributes in the destination datset.

flushInput

protected void flushInput()

This will remove all buffered instances from the inputformat dataset. Use this method rather than getInputFormat().delete();

inputFormat

public boolean inputFormat(Instances instanceInfo)
                    throws java.lang.Exception

Deprecated. use setInputFormat(Instances) instead.

Throws:

java.lang.Exception

setInputFormat

public boolean setInputFormat(Instances instanceInfo)
                       throws java.lang.Exception

Sets the format of the input instances. If the filter is able to determine the output format before seeing any input instances, it does so here. This default implementation clears the output format and output queue, and the new batch flag is set. Overriders should call super.setInputFormat(Instances)

Parameters:

instanceInfo - an Instances object containing the input instance structure (any instances contained in the object are ignored - only the structure is required).

Returns:

true if the outputFormat may be collected immediately

Throws:

java.lang.Exception - if the inputFormat can't be set successfully

outputFormat

public Instances outputFormat()

Deprecated. use getOutputFormat() instead.

getOutputFormat

public Instances getOutputFormat()

Gets the format of the output instances. This should only be called after input() or batchFinished() has returned true. The relation name of the output instances should be changed to reflect the action of the filter (eg: add the filter name and options).

Returns:

an Instances object containing the output instance structure only.

Throws:

java.lang.NullPointerException - if no input structure has been defined (or the output format hasn't been determined yet)

input

public boolean input(Instance instance)
              throws java.lang.Exception

Input an instance for filtering. Ordinarily the instance is processed and made available for output immediately. Some filters require all instances be read before producing output, in which case output instances should be collected after calling batchFinished(). If the input marks the start of a new batch, the output queue is cleared. This default implementation assumes all instance conversion will occur when batchFinished() is called.

Parameters:

instance - the input instance

Returns:

true if the filtered instance may now be collected with output().

Throws:

java.lang.NullPointerException - if the input format has not been defined.

java.lang.Exception - if the input instance was not of the correct format or if there was a problem with the filtering.

batchFinished

public boolean batchFinished()
                      throws java.lang.Exception

Signify that this batch of input to the filter is finished. If the filter requires all instances prior to filtering, output() may now be called to retrieve the filtered instances. Any subsequent instances filtered should be filtered based on setting obtained from the first batch (unless the inputFormat has been re-assigned or new options have been set). This default implementation assumes all instance processing occurs during inputFormat() and input().

Returns:

true if there are instances pending output

Throws:

java.lang.NullPointerException - if no input structure has been defined,

java.lang.Exception - if there was a problem finishing the batch.

output

public Instance output()

Output an instance after filtering and remove from the output queue.

Returns:

the instance that has most recently been filtered (or null if the queue is empty).

Throws:

java.lang.NullPointerException - if no output structure has been defined

outputPeek

public Instance outputPeek()

Output an instance after filtering but do not remove from the output queue.

Returns:

the instance that has most recently been filtered (or null if the queue is empty).

Throws:

java.lang.NullPointerException - if no input structure has been defined

numPendingOutput

public int numPendingOutput()

Returns the number of instances pending output

Returns:

the number of instances pending output

Throws:

java.lang.NullPointerException - if no input structure has been defined

isOutputFormatDefined

public boolean isOutputFormatDefined()

Returns whether the output format is ready to be collected

Returns:

true if the output format is set

getStringIndices

protected int[] getStringIndices(Instances insts)

Gets an array containing the indices of all string attributes.

Parameters:

insts - the Instances to scan for string attributes.

Returns:

an array containing the indices of string attributes in the input structure. Will be zero-length if there are no string attributes

useFilter

public static Instances useFilter(Instances data,
                                  Filter filter)
                           throws java.lang.Exception

Filters an entire set of instances through a filter and returns the new set.

Parameters:

data - the data to be filtered

filter - the filter to be used

Returns:

the filtered set of data

Throws:

java.lang.Exception - if the filter can't be used successfully

filterFile

public static void filterFile(Filter filter,
                              java.lang.String[] options)
                       throws java.lang.Exception

Method for testing filters.

Throws:

java.lang.Exception - if something goes wrong or the user requests help on command options

batchFilterFile

public static void batchFilterFile(Filter filter,
                                   java.lang.String[] options)
                            throws java.lang.Exception

Method for testing filters ability to process multiple batches.

Throws:

java.lang.Exception - if something goes wrong or the user requests help on command options

main

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

Main method for testing this class.