org.knime.base.data.normalize
Class Normalizer

java.lang.Object
  org.knime.base.data.normalize.Normalizer

public final class Normalizer
extends Object

A wrapper table to normalize all DataRows. Three methods of normalization are available:

• Min-Max Normalization
• Z-Score Normalization
• Normalization by decimal scaling

Important ! Be sure to pull a new DataTableSpec with generateNewSpec(DataTableSpec, String[]), because IntCell columns are converted to DoubleCell columns.

Author:

Nicolas Cebron, University of Konstanz

Constructor Summary

Normalizer(BufferedDataTable table, String[] columns)
Prepares a Normalizer to process the buffered data table table.

Normalizer(StatisticsTable table, String[] columns)
Prepares a Normalizer to process the StatisticsTable table (actually no traversing is done here).

Method Summary

AffineTransTable	doDecimalScaling(ExecutionMonitor exec) Does the decimal scaling.
AffineTransTable	doMinMaxNorm(double newmax, double newmin, ExecutionMonitor exec) Does the Min-Max Normalization.
AffineTransTable	doZScoreNorm(ExecutionMonitor exec) Does the Z-Score Normalization.
static DataTableSpec	generateNewSpec(DataTableSpec inspec, String[] columns) Creates a new DataTableSpec.
String	getErrorMessage()
(package private) void	setErrorMessage(String message) Sets an error message, if something went wrong during initialization.

Methods inherited from class java.lang.Object

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

Constructor Detail

Normalizer

public Normalizer(BufferedDataTable table,
                  String[] columns)

Prepares a Normalizer to process the buffered data table table. Only columns as contained in the array argument are considered.

Parameters:

table - table to be wrapped

columns - to work on

See Also:

DataTable.getDataTableSpec()

Normalizer

public Normalizer(StatisticsTable table,
                  String[] columns)

Prepares a Normalizer to process the StatisticsTable table (actually no traversing is done here).

Parameters:

table - table to be wrapped

columns - to work on

See Also:

DataTable.getDataTableSpec()

Method Detail

generateNewSpec

public static final DataTableSpec generateNewSpec(DataTableSpec inspec,
                                                  String[] columns)

Creates a new DataTableSpec. IntCell-columns are converted to DoubleCell-columns.

Parameters:

inspec - the DataTableSpec of the input table

columns - the columns that are normalized

Returns:

DataTableSpec for the output table

doMinMaxNorm

public AffineTransTable doMinMaxNorm(double newmax,
                                     double newmin,
                                     ExecutionMonitor exec)
                              throws CanceledExecutionException

Does the Min-Max Normalization.

Parameters:

newmax - the new maximum

newmin - the new minimum

exec - an object to check for user cancelations. Can be null.

Returns:

normalized DataTable

Throws:

CanceledExecutionException - if user canceled

doZScoreNorm

public AffineTransTable doZScoreNorm(ExecutionMonitor exec)
                              throws CanceledExecutionException

Does the Z-Score Normalization.

Parameters:

exec - an object to check for user cancelations. Can be null.

Returns:

the normalized DataTable

Throws:

CanceledExecutionException - if user canceled

doDecimalScaling

public AffineTransTable doDecimalScaling(ExecutionMonitor exec)
                                  throws CanceledExecutionException

Does the decimal scaling.

Parameters:

exec - an object to check for user cancelations. Can be null.

Returns:

the normalized DataTable

Throws:

CanceledExecutionException - if user canceled

setErrorMessage

void setErrorMessage(String message)

Sets an error message, if something went wrong during initialization.

Parameters:

message - the message to set.

getErrorMessage

public String getErrorMessage()

Returns:

error message if something went wrong, null otherwise.