weka.distributed

Class CSVToARFFHeaderMapTask

java.lang.Object

weka.distributed.CSVToARFFHeaderMapTask

All Implemented Interfaces:

java.io.Serializable, weka.core.OptionHandler

public class CSVToARFFHeaderMapTask
extends java.lang.Object
implements weka.core.OptionHandler, java.io.Serializable

A map task that processes incoming lines in CSV format and builds up header information. Can be configured with information on which columns to force to be nominal, string, date etc. Nominal values can be determined automatically or pre-supplied by the user. In addition to determining the format of the columns in the data it also can compute meta data such as means, modes, counts, standard deviations etc. These statistics get encoded in special "summary" attributes in the header file - one for each numeric or nominal attribute in the data.

Version:

$Revision: 13989 $

Author:

Mark Hall (mhall{[at]}pentaho{[dot]}com)

See Also:

Serialized Form

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	CSVToARFFHeaderMapTask.HeaderAndQuantileDataHolder Container class for a Instances header with basic summary stats and a map of TDigest quantile estimators for numeric attributes

Field Summary

Fields

Modifier and Type	Field and Description
static java.lang.String	ARFF_SUMMARY_ATTRIBUTE_PREFIX Attribute name prefix for a summary statistics attribute
static int	MAX_PARSING_ERRORS

Constructor Summary

Constructors

Constructor and Description
CSVToARFFHeaderMapTask() Constructor
CSVToARFFHeaderMapTask(boolean suppressQuantileOptions) Constructor
CSVToARFFHeaderMapTask(boolean suppressQuantileOptions, boolean suppressCSVParsingOptions) Constructor

Method Summary

Modifier and Type	Method and Description
static CSVToARFFHeaderMapTask	combine(java.util.List<CSVToARFFHeaderMapTask> tasks) Performs a "combine" operation using the supplied partial CSVToARFFHeaderMapTask tasks.
java.lang.String	compressionLevelForQuartileEstimationTipText() Returns the tip text for this property.
java.lang.String	computeQuartilesAsPartOfSummaryStatsTipText() Returns the tip text for this property.
java.lang.String	dateAttributesTipText() Returns the tip text for this property.
java.lang.String	dateFormatTipText() Returns the tip text for this property.
void	deSerializeAllQuantileEstimators() Deserialize all TDigest quantile estimators in use
java.lang.String	enclosureCharactersTipText() Returns the tip text for this property.
java.lang.String	fieldSeparatorTipText() Returns the tip text for this property.
void	fromHeader(weka.core.Instances headerWithSummary, java.util.Map<java.lang.String,TDigest> quantileEstimators) Initialize internal state using the supplied ARFF header with summary attributes.
void	generateNames(int numAtts) Generate attribute names.
void	generateNames(int initial, int numAtts) Generate attribute names.
double	getCompressionLevelForQuartileEstimation() Get the compression level to use in the TDigest quantile estimators
boolean	getComputeQuartilesAsPartOfSummaryStats() Get whether to include estimated quartiles in the profiling stats
java.lang.String	getDateAttributes() Returns the current attribute range to be forced to type date.
java.lang.String	getDateFormat() Get the format to use for parsing date values.
java.lang.String	getDefaultValue(int attIndex) Get the default label for a given attribute.
java.lang.String	getEnclosureCharacters() Get the character(s) to use/recognize as string enclosures
java.lang.String	getFieldSeparator() Returns the character used as column separator.
weka.core.Instances	getHeader() get the header information (as an Instances object) from what has been seen so far by this map task
weka.core.Instances	getHeader(int numFields, java.util.List<java.lang.String> attNames) Get a header constructed using the supplied attribute names.
CSVToARFFHeaderMapTask.HeaderAndQuantileDataHolder	getHeaderAndQuantileEstimators() Get the header information and the encoded quantile estimators
java.lang.String	getMissingValue() Returns the current placeholder for missing values.
java.lang.String	getNominalAttributes() Returns the current attribute range to be forced to type nominal.
java.lang.Object[]	getNominalDefaultLabelSpecs() Get the default label specifications for nominal attributes
java.lang.Object[]	getNominalLabelSpecs() Get label specifications for nominal attributes.
int	getNumDecimalPlaces() Get the number of decimal places for outputting summary stats
java.lang.String[]	getOptions()
java.lang.String	getStringAttributes() Returns the current attribute range to be forced to type string.
boolean	getTreatUnparsableNumericValuesAsMissing() Get whether, for hitherto thought to be numeric columns, to treat any unparsable values as missing value.
boolean	getTreatZerosAsMissing() Get whether to treat zeros as missing values for numeric attributes when computing summary statistics.
boolean	headerAvailableImmediately(int numFields, java.util.List<java.lang.String> attNames, java.lang.StringBuffer problems) Check if the header can be produced immediately without having to do a pre-processing pass to determine and unify nominal attribute values.
void	initParserOnly(java.util.List<java.lang.String> attNames) Only initialize enough stuff in order to parse rows and construct instances
static java.util.List<java.lang.String>	instanceHeaderToAttributeNameList(weka.core.Instances header)
java.util.Enumeration<weka.core.Option>	listOptions()
static void	main(java.lang.String[] args)
weka.core.Instance	makeInstance(weka.core.Instances trainingHeader, boolean setStringValues, java.lang.String[] parsed) Utility method for Constructing a dense instance given an array of parsed CSV values
weka.core.Instance	makeInstance(weka.core.Instances trainingHeader, boolean setStringValues, java.lang.String[] parsed, boolean sparse) Utility method for Constructing an instance given an array of parsed CSV values
weka.core.Instance	makeInstanceFromObjectRow(weka.core.Instances trainingHeader, boolean setStringValues, java.lang.Object[] row, boolean sparse) Utility method for Constructing an instance given an array of Objects
java.lang.String	missingValueTipText() Returns the tip text for this property.
java.lang.String	nominalAttributesTipText() Returns the tip text for this property.
java.lang.String	nominalDefaultLabelSpecsTipText() Returns the tip text for this property.
java.lang.String	nominalLabelSpecsTipText() Returns the tip text for this property.
java.lang.String[]	parseRowOnly(java.lang.String row) Just parse a row.
void	processRow(java.lang.String row, java.util.List<java.lang.String> attNames) Process a row of data coming into the map.
void	processRowValues(java.lang.Object[] fieldVals, java.util.List<java.lang.String> attNames) Process a tokenized row of values.
void	serializeAllQuantileEstimators() Serialize all TDigest quantile estimators in use
void	setCompressionLevelForQuartileEstimation(double compression) Set the compression level to use in the TDigest quantile estimators
void	setComputeQuartilesAsPartOfSummaryStats(boolean c) Set whether to include estimated quartiles in the profiling stats
void	setDateAttributes(java.lang.String value) Set the attribute range to be forced to type date.
void	setDateFormat(java.lang.String value) Set the format to use for parsing date values.
void	setEnclosureCharacters(java.lang.String enclosure) Set the character(s) to use/recognize as string enclosures
void	setFieldSeparator(java.lang.String value) Sets the character used as column separator.
void	setMissingValue(java.lang.String value) Sets the placeholder for missing values.
void	setNominalAttributes(java.lang.String value) Sets the attribute range to be forced to type nominal.
void	setNominalDefaultLabelSpecs(java.lang.Object[] specs) Set the default label specifications for nominal attributes
void	setNominalLabelSpecs(java.lang.Object[] specs) Set label specifications for nominal attributes.
void	setNumDecimalPlaces(int numDecimalPlaces) Set the number of decimal places for outputting summary stats
void	setOptions(java.lang.String[] options)
void	setStringAttributes(java.lang.String value) Sets the attribute range to be forced to type string.
void	setTreatUnparsableNumericValuesAsMissing(boolean unparsableNumericValuesToMissing) Set whether, for hitherto thought to be numeric columns, to treat any unparsable values as missing value.
void	setTreatZerosAsMissing(boolean t) Set whether to treat zeros as missing values for numeric attributes when computing summary statistics.
java.lang.String	stringAttributesTipText() Returns the tip text for this property.
static void	updateSummaryStats(java.util.Map<java.lang.String,Stats> summaryStats, java.util.Map<java.lang.String,StringStats> backupStringStats, java.lang.String attName, double value, java.lang.String nominalLabel, boolean isNominal, boolean isString, boolean treatZeroAsMissing, boolean estimateQuantiles, double quantileCompression) Update the summary statistics for a given attribute with the given value

Methods inherited from class java.lang.Object

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

Methods inherited from interface weka.core.OptionHandler

makeCopy

Field Detail

ARFF_SUMMARY_ATTRIBUTE_PREFIX

public static final java.lang.String ARFF_SUMMARY_ATTRIBUTE_PREFIX

Attribute name prefix for a summary statistics attribute

See Also:

Constant Field Values

MAX_PARSING_ERRORS

public static final int MAX_PARSING_ERRORS

See Also:

Constant Field Values

Constructor Detail

CSVToARFFHeaderMapTask

public CSVToARFFHeaderMapTask()

Constructor

CSVToARFFHeaderMapTask

public CSVToARFFHeaderMapTask(boolean suppressQuantileOptions)

Constructor

Parameters:

suppressQuantileOptions - true if commandline options relating to quantile estimation are to be suppressed

CSVToARFFHeaderMapTask

public CSVToARFFHeaderMapTask(boolean suppressQuantileOptions,
                              boolean suppressCSVParsingOptions)

Constructor

Parameters:

suppressQuantileOptions - true if command line options relating to quantile estimation are to be suppressed

suppressCSVParsingOptions - true if command line options relating to CSV parsing are to be suppressed

Method Detail

updateSummaryStats

public static void updateSummaryStats(java.util.Map<java.lang.String,Stats> summaryStats,
                                      java.util.Map<java.lang.String,StringStats> backupStringStats,
                                      java.lang.String attName,
                                      double value,
                                      java.lang.String nominalLabel,
                                      boolean isNominal,
                                      boolean isString,
                                      boolean treatZeroAsMissing,
                                      boolean estimateQuantiles,
                                      double quantileCompression)

Update the summary statistics for a given attribute with the given value

Parameters:

summaryStats - the map of summary statistics

backupStringStats - the temporary map of backup string stats kept for numeric fields (this can be null in cases where we are sure that there is no chance of unparsable numeric values occuring)

attName - the name of the attribute being updated

value - the value to update with (if the attribute is numeric)

nominalLabel - holds the label/string for the attribute (if it is nominal or string)

isNominal - true if the attribute is nominal

isString - true if the attribute is a string attribute

treatZeroAsMissing - treats zero as missing value for numeric attributes

estimateQuantiles - true if we should estimate quantiles too

quantileCompression - the compression level to use in the TDigest estimators

instanceHeaderToAttributeNameList

public static java.util.List<java.lang.String> instanceHeaderToAttributeNameList(weka.core.Instances header)

main

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

combine

public static CSVToARFFHeaderMapTask combine(java.util.List<CSVToARFFHeaderMapTask> tasks)
                                      throws DistributedWekaException

Performs a "combine" operation using the supplied partial CSVToARFFHeaderMapTask tasks. This is essentially a reduce operation, but returns a single CSVToARFFHeaderMapTask object (rather than the final header that is produced by CSVToARFFHeaderReduceTask). This allows several reduce stages to be implemented (if desired) or partial reduces to occur in parallel.

Parameters:

tasks - a list of CSVToARFFHeaderMapTasks to "combine"

Returns:

a CSVToARFFHeaderMapTask with the merged state

Throws:

DistributedWekaException - if a problem occurs

listOptions

public java.util.Enumeration<weka.core.Option> listOptions()

Specified by:

listOptions in interface weka.core.OptionHandler

getOptions

public java.lang.String[] getOptions()

Specified by:

getOptions in interface weka.core.OptionHandler

setOptions

public void setOptions(java.lang.String[] options)
                throws java.lang.Exception

Specified by:

setOptions in interface weka.core.OptionHandler

Throws:

java.lang.Exception

setNumDecimalPlaces

public void setNumDecimalPlaces(int numDecimalPlaces)

Set the number of decimal places for outputting summary stats

Parameters:

numDecimalPlaces - number of decimal places to use

getNumDecimalPlaces

public int getNumDecimalPlaces()

Get the number of decimal places for outputting summary stats

Returns:

number of decimal places to use

setTreatUnparsableNumericValuesAsMissing

public void setTreatUnparsableNumericValuesAsMissing(boolean unparsableNumericValuesToMissing)

Set whether, for hitherto thought to be numeric columns, to treat any unparsable values as missing value.

Parameters:

unparsableNumericValuesToMissing -

getTreatUnparsableNumericValuesAsMissing

public boolean getTreatUnparsableNumericValuesAsMissing()

Get whether, for hitherto thought to be numeric columns, to treat any unparsable values as missing value.

Returns:

true if unparsable numeric values are to be treated as missing

getTreatZerosAsMissing

public boolean getTreatZerosAsMissing()

Get whether to treat zeros as missing values for numeric attributes when computing summary statistics.

Returns:

true if zeros are to be treated as missing values for the purposes of computing summary stats.

setTreatZerosAsMissing

public void setTreatZerosAsMissing(boolean t)

Set whether to treat zeros as missing values for numeric attributes when computing summary statistics.

Parameters:

t - true if zeros are to be treated as missing values for the purposes of computing summary stats.

getCompressionLevelForQuartileEstimation

public double getCompressionLevelForQuartileEstimation()

Get the compression level to use in the TDigest quantile estimators

Returns:

the compression level (smaller values give higher compression and less accurate estimates).

setCompressionLevelForQuartileEstimation

public void setCompressionLevelForQuartileEstimation(double compression)

Set the compression level to use in the TDigest quantile estimators

Parameters:

compression - the compression level (smaller values give higher compression and less accurate estimates).

compressionLevelForQuartileEstimationTipText

public java.lang.String compressionLevelForQuartileEstimationTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the explorer/experimenter gui

getComputeQuartilesAsPartOfSummaryStats

public boolean getComputeQuartilesAsPartOfSummaryStats()

Get whether to include estimated quartiles in the profiling stats

Returns:

true if quartiles are to be estimated

setComputeQuartilesAsPartOfSummaryStats

public void setComputeQuartilesAsPartOfSummaryStats(boolean c)

Set whether to include estimated quartiles in the profiling stats

Parameters:

c - true if quartiles are to be estimated

computeQuartilesAsPartOfSummaryStatsTipText

public java.lang.String computeQuartilesAsPartOfSummaryStatsTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the explorer/experimenter gui

getMissingValue

public java.lang.String getMissingValue()

Returns the current placeholder for missing values.

Returns:

the placeholder

setMissingValue

public void setMissingValue(java.lang.String value)

Sets the placeholder for missing values.

Parameters:

value - the placeholder

missingValueTipText

public java.lang.String missingValueTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the explorer/experimenter gui

getStringAttributes

public java.lang.String getStringAttributes()

Returns the current attribute range to be forced to type string.

Returns:

the range

setStringAttributes

public void setStringAttributes(java.lang.String value)

Sets the attribute range to be forced to type string.

Parameters:

value - the range

stringAttributesTipText

public java.lang.String stringAttributesTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the explorer/experimenter gui

getNominalAttributes

public java.lang.String getNominalAttributes()

Returns the current attribute range to be forced to type nominal.

Returns:

the range

setNominalAttributes

public void setNominalAttributes(java.lang.String value)

Sets the attribute range to be forced to type nominal.

Parameters:

value - the range

nominalAttributesTipText

public java.lang.String nominalAttributesTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the explorer/experimenter gui

getDateFormat

public java.lang.String getDateFormat()

Get the format to use for parsing date values.

Returns:

the format to use for parsing date values.

setDateFormat

public void setDateFormat(java.lang.String value)

Set the format to use for parsing date values.

Parameters:

value - the format to use.

dateFormatTipText

public java.lang.String dateFormatTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the explorer/experimenter gui

getDateAttributes

public java.lang.String getDateAttributes()

Returns the current attribute range to be forced to type date.

Returns:

the range.

setDateAttributes

public void setDateAttributes(java.lang.String value)

Set the attribute range to be forced to type date.

Parameters:

value - the range

dateAttributesTipText

public java.lang.String dateAttributesTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the explorer/experimenter gui

enclosureCharactersTipText

public java.lang.String enclosureCharactersTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the explorer/experimenter gui

getEnclosureCharacters

public java.lang.String getEnclosureCharacters()

Get the character(s) to use/recognize as string enclosures

Returns:

the characters to use as string enclosures

setEnclosureCharacters

public void setEnclosureCharacters(java.lang.String enclosure)

Set the character(s) to use/recognize as string enclosures

Parameters:

enclosure - the characters to use as string enclosures

getFieldSeparator

public java.lang.String getFieldSeparator()

Returns the character used as column separator.

Returns:

the character to use

setFieldSeparator

public void setFieldSeparator(java.lang.String value)

Sets the character used as column separator.

Parameters:

value - the character to use

fieldSeparatorTipText

public java.lang.String fieldSeparatorTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the explorer/experimenter gui

nominalDefaultLabelSpecsTipText

public java.lang.String nominalDefaultLabelSpecsTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the explorer/experimenter gui

getNominalDefaultLabelSpecs

public java.lang.Object[] getNominalDefaultLabelSpecs()

Get the default label specifications for nominal attributes

Returns:

an array of default label specifications

setNominalDefaultLabelSpecs

public void setNominalDefaultLabelSpecs(java.lang.Object[] specs)

Set the default label specifications for nominal attributes

Parameters:

specs - an array of default label specifications

nominalLabelSpecsTipText

public java.lang.String nominalLabelSpecsTipText()

Returns the tip text for this property.

Returns:

tip text for this property suitable for displaying in the explorer/experimenter gui

getNominalLabelSpecs

public java.lang.Object[] getNominalLabelSpecs()

Get label specifications for nominal attributes.

Returns:

an array of label specifications

setNominalLabelSpecs

public void setNominalLabelSpecs(java.lang.Object[] specs)

Set label specifications for nominal attributes.

Parameters:

specs - an array of label specifications

generateNames

public void generateNames(int initial,
                          int numAtts)

Generate attribute names. Attributes are named "attinitial", "attinitial+1", ..., "attinitial+numAtts-1"

Parameters:

initial - the number to use for the first attribute

numAtts - the number of attributes to generate

generateNames

public void generateNames(int numAtts)

Generate attribute names. Attributes are named "att0", "att1", ... "attnumAtts-1"

Parameters:

numAtts - the number of attribute names to generate

initParserOnly

public void initParserOnly(java.util.List<java.lang.String> attNames)

Only initialize enough stuff in order to parse rows and construct instances

Parameters:

attNames - the names of the attributes to use

parseRowOnly

public java.lang.String[] parseRowOnly(java.lang.String row)
                                throws java.io.IOException

Just parse a row.

Parameters:

row - the row to parse

Returns:

the values of the row in an array

Throws:

java.io.IOException - if a problem occurs

processRowValues

public void processRowValues(java.lang.Object[] fieldVals,
                             java.util.List<java.lang.String> attNames)
                      throws DistributedWekaException,
                             java.io.IOException

Process a tokenized row of values. attNames may be non-null for the first row and is optional. If not supplied then names will be generated on receiving the first row of data. An exception will be raised on subsequent rows that don't have the same number of fields as seen in the first row

Parameters:

fieldVals - the row values to process

attNames - the names of the attributes (fields)

Throws:

if - the number of fields in the current row does not match the number of attribute names

DistributedWekaException

java.io.IOException

processRow

public void processRow(java.lang.String row,
                       java.util.List<java.lang.String> attNames)
                throws DistributedWekaException,
                       java.io.IOException

Process a row of data coming into the map. Split the row into fields and initialize if this is the first row seen. attNames may be non-null for the first row and is optional. If not supplied then names will be generated on receiving the first row of data. An exception will be raised on subsequent rows that don't have the same number of fields as seen in the first row

Parameters:

row - the row to process

attNames - the names of the attributes (fields)

Throws:

if - the number of fields in the current row does not match the number of attribute names

DistributedWekaException

java.io.IOException

getHeader

public weka.core.Instances getHeader()

get the header information (as an Instances object) from what has been seen so far by this map task

Returns:

the header information as an Instances object

getHeaderAndQuantileEstimators

public CSVToARFFHeaderMapTask.HeaderAndQuantileDataHolder getHeaderAndQuantileEstimators()
                                                                                  throws DistributedWekaException

Get the header information and the encoded quantile estimators

Returns:

a holder instance containing both the header information and encoded quantile estimators

Throws:

DistributedWekaException - if we are not computing summary statistics or we are computing statistics but not quantiles

serializeAllQuantileEstimators

public void serializeAllQuantileEstimators()

Serialize all TDigest quantile estimators in use

deSerializeAllQuantileEstimators

public void deSerializeAllQuantileEstimators()

Deserialize all TDigest quantile estimators in use

headerAvailableImmediately

public boolean headerAvailableImmediately(int numFields,
                                          java.util.List<java.lang.String> attNames,
                                          java.lang.StringBuffer problems)

Check if the header can be produced immediately without having to do a pre-processing pass to determine and unify nominal attribute values. All types should be specified via the ranges and nominal label specs.

Parameters:

numFields - number of fields in the data

attNames - the names of the attributes (in order)

problems - a StringBuffer to hold problem descriptions (if any)

Returns:

true if the header can be generated immediately with out a pre-processing job

getHeader

public weka.core.Instances getHeader(int numFields,
                                     java.util.List<java.lang.String> attNames)
                              throws DistributedWekaException

Get a header constructed using the supplied attribute names. This should only be called in the situation where the data does not require a pre-processing pass to determine and unify nominal attribute values. All types should be specified via the ranges and nominal label specifications.

Parameters:

numFields - the number of attributes in the data

attNames - the attribute names to use. May be null, in which case names are generated

Returns:

an Instances object encapsulating header information

Throws:

DistributedWekaException - if nominal attributes have been specified but there are one or more tha have no user-supplied label specifications

fromHeader

public void fromHeader(weka.core.Instances headerWithSummary,
                       java.util.Map<java.lang.String,TDigest> quantileEstimators)
                throws DistributedWekaException

Initialize internal state using the supplied ARFF header with summary attributes. Assumes that setOptions() has already been called on this instance of CSVToARFFHeaderMapTask.

Parameters:

headerWithSummary - the ARFF header (with summary attributes) to initialize with

quantileEstimators - a map (keyed by attribute name) of TDigest estimators for numeric attributes (can be null if quantiles are not being estimated)

Throws:

DistributedWekaException - if a problem occurs

makeInstance

public weka.core.Instance makeInstance(weka.core.Instances trainingHeader,
                                       boolean setStringValues,
                                       java.lang.String[] parsed)
                                throws java.lang.Exception

Utility method for Constructing a dense instance given an array of parsed CSV values

Parameters:

trainingHeader - the header to associate the instance with. Does not add the new instance to this data set; just gives the instance a reference to the header

setStringValues - true if any string values should be set in the header as opposed to being added to the header (i.e. accumulating in the header).

parsed - the array of parsed CSV values

Returns:

an Instance

Throws:

java.lang.Exception - if a problem occurs

makeInstance

public weka.core.Instance makeInstance(weka.core.Instances trainingHeader,
                                       boolean setStringValues,
                                       java.lang.String[] parsed,
                                       boolean sparse)
                                throws java.lang.Exception

Utility method for Constructing an instance given an array of parsed CSV values

Parameters:

trainingHeader - the header to associate the instance with. Does not add the new instance to this data set; just gives the instance a reference to the header

setStringValues - true if any string values should be set in the header as opposed to being added to the header (i.e. accumulating in the header).

parsed - the array of parsed CSV values

sparse - true if the new instance is to be a sparse instance

Returns:

an Instance

Throws:

java.lang.Exception - if a problem occurs

makeInstanceFromObjectRow

public weka.core.Instance makeInstanceFromObjectRow(weka.core.Instances trainingHeader,
                                                    boolean setStringValues,
                                                    java.lang.Object[] row,
                                                    boolean sparse)
                                             throws java.lang.Exception

Utility method for Constructing an instance given an array of Objects

Parameters:

trainingHeader - the header to associate the instance with. Does not add the new instance to this data set; just gives the instance a reference to the header

setStringValues - true if any string values should be set in the header as opposed to being added to the header (i.e. accumulating in the header).

row - the array of Object values

sparse - true if the new instance is to be a sparse instance

Returns:

an Instance

Throws:

java.lang.Exception - if a problem occurs

getDefaultValue

public java.lang.String getDefaultValue(int attIndex)

Get the default label for a given attribute. May be null if a default value hasn't been specified

Parameters:

attIndex - the index (0-based) of the attribute to get the default value for

Returns:

the default value or null (if a default has not been specified)