Catalano.Statistics

Class Normalizations

java.lang.Object

Catalano.Statistics.Normalizations

public class Normalizations
extends java.lang.Object

Data preprocessing describes any type of processing performed on raw data to prepare it for another processing procedure.

Method Summary

Modifier and Type	Method and Description
static double[]	DecimalScaling(double[] data) Decimal scaling.
static double[]	MinMaxNormalization(double[] data, double min, double max) Map the data to a specified range, Smin to Smax, but still retain the relationship between the values.
static double[]	RangeNormalization(double[] data, double fromMin, double fromMax, double toMin, double toMax) Converts the value x (which is measured in the scale 'from') to another value measured in the scale 'to'.
static double[]	SoftmaxScaling(double[] data, double r) Nonlinear method may be desired if the data distribution is skewed.
static double[]	StandartNormalDensity(double[] data) A commonly used statistical-based method to normalize these measures is to take each vector component and subtract the mean and divide by the standard deviation.
static double[]	UnitVectorNormalization(double[] data) Will modify the feature vectors so that they all have a magnitude of 1.

Methods inherited from class java.lang.Object

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

Method Detail

DecimalScaling

public static double[] DecimalScaling(double[] data)

Decimal scaling.

Parameters:

data - Data.

Returns:

Scaled data.

RangeNormalization

public static double[] RangeNormalization(double[] data,
                                          double fromMin,
                                          double fromMax,
                                          double toMin,
                                          double toMax)

Converts the value x (which is measured in the scale 'from') to another value measured in the scale 'to'.

Parameters:

data - Data.

fromMin - Range min from.

fromMax - Range max from.

toMin - Range min from.

toMax - Range max from.

Returns:

Normalized data.

UnitVectorNormalization

public static double[] UnitVectorNormalization(double[] data)

Will modify the feature vectors so that they all have a magnitude of 1. If this is one we will retain only directional information about the vector, which preserves relationships between the features, but loses magnitudes.

Parameters:

data - Data.

Returns:

Normalized data.

StandartNormalDensity

public static double[] StandartNormalDensity(double[] data)

A commonly used statistical-based method to normalize these measures is to take each vector component and subtract the mean and divide by the standard deviation.

Parameters:

data - Data.

Returns:

Normalized data.

MinMaxNormalization

public static double[] MinMaxNormalization(double[] data,
                                           double min,
                                           double max)

Map the data to a specified range, Smin to Smax, but still retain the relationship between the values.

Parameters:

data - Data.

min - Range min.

max - Range max.

Returns:

Normalized data.

SoftmaxScaling

public static double[] SoftmaxScaling(double[] data,
                                      double r)

Nonlinear method may be desired if the data distribution is skewed. This is essentially a method that compresses the data into the range 0 to 1.

Parameters:

data - Data.

r - Determines the range of values for the feature, that will fall into the linear range.

Returns:

Normalized data.