edu.udo.cs.yale.example
Class AbstractExampleTable

java.lang.Object
  edu.udo.cs.yale.example.AbstractExampleTable

All Implemented Interfaces:

ExampleTable

Direct Known Subclasses:

DatabaseExampleTable, MemoryExampleTable, RandomExampleTable

public abstract class AbstractExampleTable
extends java.lang.Object
implements ExampleTable

This class is the core data supplier for example sets. Several example sets can use the same data and access the attribute values by reference.

Version:

$Id: AbstractExampleTable.java,v 2.3 2006/08/23 08:13:53 ingomierswa Exp $

Author:

Simon Fischer, Ingo Mierswa

Field Summary

private java.util.List<Attribute>	attributes List of instances of Attribute.
private int	currentlyHighestBlockNr
private java.util.List<java.lang.Integer>	unusedColumnList List of Integers referencing indices of columns that were removed, e.g.

Constructor Summary

AbstractExampleTable(java.util.List<Attribute> attributes)
Creates a new ExampleTable.

Method Summary

int	addAttribute(Attribute a) Adds the attribute to the list of attributes assigning it a free column index.
void	addAttributes(java.util.Collection<Attribute> newAttributes) Adds all Attributes in newAttributes to the end of the list of attributes, creating new data columns if necessary.
ExampleSet	createCompleteExampleSet(Attribute label) Returns a new example set with all attributes switched on.
ExampleSet	createCompleteExampleSet(Attribute label, Attribute predictedLabel, Attribute weight, Attribute idAttribute) Returns a new example set with all attributes switched on.
ExampleSet	createCompleteExampleSet(java.util.Map<java.lang.String,Attribute> specialAttributes) Returns a new example set with all attributes switched on.
ExampleSet	createExampleSet(AttributeSet attributeSet) Returns a new example set with all attributes of the given attribute set.
Attribute	findAttribute(java.lang.String name) Returns the attribute with the given name.
Attribute	getAttribute(Attribute attribute) Returns an Attribute which is constructed in the same way as attribute.
Attribute	getAttribute(int i) Returns the attribute of the column number i.
int	getAttributeCount() Returns the number of non null attributes.
Attribute[]	getAttributes() Returns a new array containing all Attributes.
int	getBlockEndIndex(int startindex) Returns the last attribute index belonging to the block starting at startindex.
abstract DataRowReader	getDataReader() Returns an Iterator for example data given as DataRow objects.
abstract DataRow	getDataRow(int index) Returns the i-th data row.
int	getHighestBlockNr() Gets the currently highest used block number.
int	getNextFreeBlockNr() Returns a new unused blockNr.
int	getNumberOfAttributes() Returns the number of attributes.
abstract int	getSize() Returns the number of examples.
void	removeAttribute(Attribute attribute) Equivalent to calling removeAttribute(attribute.getTableIndex()).
void	removeAttribute(int index) Sets the attribute with the given index to null.
java.lang.String	toDataString()
java.lang.String	toString()

Methods inherited from class java.lang.Object

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

Field Detail

currentlyHighestBlockNr

private int currentlyHighestBlockNr

attributes

private java.util.List<Attribute> attributes

List of instances of Attribute. The i-th entry in the list belongs to the i-th data column. Whenever attributes are removed from the list of attributes (e.g. they were intermediate predicted labels used only within a validation chain), the succeeding entries do not move up, but the entry is replaced by a null entry and this index is added to unusedColumnList as an Integer.

unusedColumnList

private java.util.List<java.lang.Integer> unusedColumnList

List of Integers referencing indices of columns that were removed, e.g. predicted labels that were used within a validation chain but are not needed any longer. Any of the columns in this list may be used when a new attribute is created. The list is used as a queue.

Constructor Detail

AbstractExampleTable

public AbstractExampleTable(java.util.List<Attribute> attributes)

Creates a new ExampleTable.

Parameters:

attributes - List of Attribute. The indices of the attibutes are set to values reflecting their position in the list.

Method Detail

getSize

public abstract int getSize()

Returns the number of examples.

Specified by:

getSize in interface ExampleTable

getDataReader

public abstract DataRowReader getDataReader()

Returns an Iterator for example data given as DataRow objects. This should be used in all cases where iteration is desired. Since getDataRow(int) does not ensure to work in an efficient way the usage of this method is preferred (instead using for-loops).

Specified by:

getDataReader in interface ExampleTable

getDataRow

public abstract DataRow getDataRow(int index)

Returns the i-th data row. Calling methods cannot rely on the efficiency of this method. Memory based example tables should return the data row in O(1).

Specified by:

getDataRow in interface ExampleTable

getAttributes

public Attribute[] getAttributes()

Returns a new array containing all Attributes.

Specified by:

getAttributes in interface ExampleTable

getAttribute

public Attribute getAttribute(int i)

Returns the attribute of the column number i. Attention: This value may return null if the column was marked unused.

Specified by:

getAttribute in interface ExampleTable

findAttribute

public Attribute findAttribute(java.lang.String name)
                        throws OperatorException

Returns the attribute with the given name.

Specified by:

findAttribute in interface ExampleTable

Throws:

OperatorException

addAttributes

public void addAttributes(java.util.Collection<Attribute> newAttributes)

Adds all Attributes in newAttributes to the end of the list of attributes, creating new data columns if necessary.

Specified by:

addAttributes in interface ExampleTable

addAttribute

public int addAttribute(Attribute a)

Adds the attribute to the list of attributes assigning it a free column index.

Specified by:

addAttribute in interface ExampleTable

removeAttribute

public void removeAttribute(Attribute attribute)

Equivalent to calling removeAttribute(attribute.getTableIndex()).

Specified by:

removeAttribute in interface ExampleTable

removeAttribute

public void removeAttribute(int index)

Sets the attribute with the given index to null. Afterwards, this column can be reused. Callers must make sure, that no other example set contains a reference to this column. Otherwise its data will be messed up. Ususally this is only possible if an operator generates intermediate attributes, like a validation chain or a feature generator. If the attribute already was removed, this method returns silently.

Specified by:

removeAttribute in interface ExampleTable

getNumberOfAttributes

public int getNumberOfAttributes()

Returns the number of attributes. Attention: Callers that use a for-loop and retrieving Attributes by calling getAttribute(int) must keep in mind, that some of these attributes may be null.

Specified by:

getNumberOfAttributes in interface ExampleTable

getAttributeCount

public int getAttributeCount()

Returns the number of non null attributes. Attention: Since there are null attributes in the list, the return value of this method must not be used in a for-loop!

Specified by:

getAttributeCount in interface ExampleTable

See Also:

ExampleTable.getNumberOfAttributes().

getHighestBlockNr

public int getHighestBlockNr()

Gets the currently highest used block number.

Specified by:

getHighestBlockNr in interface ExampleTable

getNextFreeBlockNr

public int getNextFreeBlockNr()

Returns a new unused blockNr. BUG: Achtung, wenn Attributdatei Blocknummer enthaelt. Wird hier nicht beruecksichtigt.

Specified by:

getNextFreeBlockNr in interface ExampleTable

getBlockEndIndex

public int getBlockEndIndex(int startindex)

Returns the last attribute index belonging to the block starting at startindex.

Specified by:

getBlockEndIndex in interface ExampleTable

getAttribute

public Attribute getAttribute(Attribute attribute)

Returns an Attribute which is constructed in the same way as attribute. Attention:

• This method do not use the equals method of attribute which only checks the name and table index (which is actually the meaning of equals) but it compares the construction descriptions of both attributes. This is especially useful for automatic feature construction algorithms which should not regenerate the same attribute.
• May return null (deleted columns).

Specified by:

getAttribute in interface ExampleTable

createExampleSet

public ExampleSet createExampleSet(AttributeSet attributeSet)

Returns a new example set with all attributes of the given attribute set.

Specified by:

createExampleSet in interface ExampleTable

createCompleteExampleSet

public ExampleSet createCompleteExampleSet(Attribute label)

Returns a new example set with all attributes switched on.

createCompleteExampleSet

public ExampleSet createCompleteExampleSet(Attribute label,
                                           Attribute predictedLabel,
                                           Attribute weight,
                                           Attribute idAttribute)

Returns a new example set with all attributes switched on.

Specified by:

createCompleteExampleSet in interface ExampleTable

createCompleteExampleSet

public ExampleSet createCompleteExampleSet(java.util.Map<java.lang.String,Attribute> specialAttributes)

Returns a new example set with all attributes switched on. The attributes in the given map will be used as special attributes, all other attributes given at creation time will be regular.

Specified by:

createCompleteExampleSet in interface ExampleTable

toString

public java.lang.String toString()

Specified by:

toString in interface ExampleTable

Overrides:

toString in class java.lang.Object

toDataString

public java.lang.String toDataString()

Specified by:

toDataString in interface ExampleTable