org.encog.ml.data.versatile

Class NormalizationHelper

java.lang.Object

org.encog.ml.data.versatile.NormalizationHelper

All Implemented Interfaces:

java.io.Serializable

public class NormalizationHelper
extends java.lang.Object
implements java.io.Serializable

This class is used to perform normalizations for methods trained with the versatile dataset.

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
NormalizationHelper()

Method Summary

Modifier and Type	Method and Description
void	addSourceColumn(ColumnDefinition def) Add a source column.
MLData	allocateInputVector() Allocate a data item large enough to hold a single input vector.
MLData	allocateInputVector(int multiplier) Allocate a data item large enough to hold several input vectors.
int	calculateNormalizedInputCount()
int	calculateNormalizedOutputCount()
void	clearInputOutput() Clear the input/output columns, but not the source columns.
void	defineMissingHandler(ColumnDefinition colDef, MissingHandler handler) Define a missing value handler.
ColumnDefinition	defineSourceColumn(java.lang.String name, int index, ColumnType colType) Define a source column.
void	defineUnknownValue(java.lang.String str) Define the string that signifies an unknown value (eg "?")
java.lang.String[]	denormalizeOutputVectorToString(MLData output) Denormalize a complete output vector to an array of strings.
CSVFormat	getFormat()
java.util.List<ColumnDefinition>	getInputColumns()
NormalizationStrategy	getNormStrategy()
java.util.List<ColumnDefinition>	getOutputColumns()
java.util.List<ColumnDefinition>	getSourceColumns()
java.util.List<java.lang.String>	getUnknownValues()
double[]	normalizeInputColumn(int i, java.lang.String value) Normalize a single input column.
void	normalizeInputVector(java.lang.String[] line, double[] data, boolean originalOrder) Normalize a string array to an input vector.
double[]	normalizeOutputColumn(int i, java.lang.String value) Normalize a single output column.
int	normalizeToVector(ColumnDefinition colDef, int outputColumn, double[] output, boolean isInput, java.lang.String value) Normalize a single column to the input vector.
double	parseDouble(java.lang.String str) Parse a double, using the correct formatter.
void	setFormat(CSVFormat format)
void	setInputColumns(java.util.List<ColumnDefinition> inputColumns)
void	setNormStrategy(NormalizationStrategy normStrategy)
void	setOutputColumns(java.util.List<ColumnDefinition> outputColumns)
void	setSourceColumns(java.util.List<ColumnDefinition> sourceColumns)
void	setStrategy(NormalizationStrategy strat) Set the normalization strategy.
java.lang.String	toString()

Methods inherited from class java.lang.Object

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

Constructor Detail

NormalizationHelper

public NormalizationHelper()

Method Detail

getSourceColumns

public java.util.List<ColumnDefinition> getSourceColumns()

Returns:

the sourceColumns

setSourceColumns

public void setSourceColumns(java.util.List<ColumnDefinition> sourceColumns)

Parameters:

sourceColumns - the sourceColumns to set

getInputColumns

public java.util.List<ColumnDefinition> getInputColumns()

Returns:

the inputColumns

setInputColumns

public void setInputColumns(java.util.List<ColumnDefinition> inputColumns)

Parameters:

inputColumns - the inputColumns to set

getOutputColumns

public java.util.List<ColumnDefinition> getOutputColumns()

Returns:

the outputColumns

setOutputColumns

public void setOutputColumns(java.util.List<ColumnDefinition> outputColumns)

Parameters:

outputColumns - the outputColumns to set

getNormStrategy

public NormalizationStrategy getNormStrategy()

Returns:

the normStrategy

setNormStrategy

public void setNormStrategy(NormalizationStrategy normStrategy)

Parameters:

normStrategy - the normStrategy to set

addSourceColumn

public void addSourceColumn(ColumnDefinition def)

Add a source column. These define the raw input.

Parameters:

def - The column definition.

defineSourceColumn

public ColumnDefinition defineSourceColumn(java.lang.String name,
                                           int index,
                                           ColumnType colType)

Define a source column. These define the raw input. Use this function if you know the index of the column in a non-header file.

Parameters:

name - The name of the column.

index - The index of the column, needed for non-headered files.

colType - The column type.

Returns:

The column definition

toString

public java.lang.String toString()

Overrides:

toString in class java.lang.Object

clearInputOutput

public void clearInputOutput()

Clear the input/output columns, but not the source columns.

normalizeInputColumn

public double[] normalizeInputColumn(int i,
                                     java.lang.String value)

Normalize a single input column.

Parameters:

i - The column definition index (from the input columns).

value - The value to normalize.

Returns:

The normalized result.

normalizeOutputColumn

public double[] normalizeOutputColumn(int i,
                                      java.lang.String value)

Normalize a single output column.

Parameters:

i - The column definition index (from the output columns).

value - The value to normalize.

Returns:

The normalized result.

calculateNormalizedInputCount

public int calculateNormalizedInputCount()

Returns:

The number of elements the input will normalize to.

calculateNormalizedOutputCount

public int calculateNormalizedOutputCount()

Returns:

The number of elements the output will normalize to.

allocateInputVector

public MLData allocateInputVector()

Allocate a data item large enough to hold a single input vector.

Returns:

The data element.

allocateInputVector

public MLData allocateInputVector(int multiplier)

Allocate a data item large enough to hold several input vectors. This is normally used for timeslices.

Parameters:

multiplier - How many input vectors.

Returns:

The data element.

denormalizeOutputVectorToString

public java.lang.String[] denormalizeOutputVectorToString(MLData output)

Denormalize a complete output vector to an array of strings.

Parameters:

output - The data vector to denorm, the source.

Returns:

The denormalized vector.

setStrategy

public void setStrategy(NormalizationStrategy strat)

Set the normalization strategy.

Parameters:

strat - The strategy.

getFormat

public CSVFormat getFormat()

Returns:

the format

setFormat

public void setFormat(CSVFormat format)

Parameters:

format - the format to set

getUnknownValues

public java.util.List<java.lang.String> getUnknownValues()

Returns:

the unknownValues

defineUnknownValue

public void defineUnknownValue(java.lang.String str)

Define the string that signifies an unknown value (eg "?")

Parameters:

str - The string for unknowns.

normalizeToVector

public int normalizeToVector(ColumnDefinition colDef,
                             int outputColumn,
                             double[] output,
                             boolean isInput,
                             java.lang.String value)

Normalize a single column to the input vector.

Parameters:

colDef - The column to normalize.

outputColumn - The current position in the vector.

output - The vector to output to.

isInput - Is this an input column.

value - The value to normalize.

Returns:

The new current position in the vector.

parseDouble

public double parseDouble(java.lang.String str)

Parse a double, using the correct formatter.

Parameters:

str - The string.

Returns:

The double.

defineMissingHandler

public void defineMissingHandler(ColumnDefinition colDef,
                                 MissingHandler handler)

Define a missing value handler.

Parameters:

colDef - The column this handler applies to.

handler - The handler.

normalizeInputVector

public void normalizeInputVector(java.lang.String[] line,
                                 double[] data,
                                 boolean originalOrder)

Normalize a string array to an input vector.

Parameters:

line - The unnormalized string array.

data - The output data.

originalOrder - Should the output be forced into the original column order?