Package org.openjump.core.attributeoperations

Class Classifier1D

java.lang.Object

org.openjump.core.attributeoperations.Classifier1D

public class Classifier1D
extends Object

Field Summary

Fields

Modifier and Type	Field	Description
static String	EQUAL_NUMBER
static String	EQUAL_RANGE
static String	JENKS_BREAKS
static String	KMEANS_OPTIMIZE
static String	MAX_BREAKS
static String	MEAN_STDEV

Constructor Summary

Constructors

Constructor	Description
Classifier1D()

Method Summary

Modifier and Type	Method	Description
static double[]	adjustLimitsKMeans(double[] data, double[] oldLimits)	Moves the limits, by assigning data points to the closest class mean value.
static double[]	calcClassMeans(double[] data, int[] classes, int numClasses)
static double	calcGVF(double[] data, double[] limits, double SDAM)	GVF (goodness of variance fit): see B.D.
static double	calcGVF(double SDAM, double SDCM)	GVF (goodness of variance fit): see B.D.
static double	calcSDAM(double[] data)	SDAM (squared deviation [from] array mean): see B.D.
static double	calcSDCM(double[] data, int[] classes, double[] classMeans, int numClasses)	SDCM (squared deviations [from] class means): see B.D.
static int[]	classifyData(double[] data, double[] limits)	Classifies the given data according to the given limits.
static double[]	classifyEqualNumber(double[] data, int numberClasses)	calculates class limits with equal number, which is euqal to the "quantiles" method.
static double[]	classifyEqualRange(double[] data, int numberClasses)	calculates class limits with equal range
static double[]	classifyKMeansOnExistingBreaks(double[] data, int numberClasses, int initialLimitAlgorithm)	calculates class limits using optimal breaks method (see e.g.
static double[]	classifyMaxBreaks(double[] data, int numberClasses)	calculates class limits using Maximum Breaks method (see e.g.
static double[]	classifyMeanStandardDeviation(double[] data, int numberClasses)	calculates class limits using mean value and standard deviation, i.e.
static double[]	classifyNaturalBreaks(double[] data, int numberClasses)	calculates class limits using Jenks's Optimisation Method(Natural Break)
static List	getAvailableClassificationMethods()
static boolean	isInClass(double val, double lowerBound, double upperBound)	Checks if value is within limits.

Methods inherited from class java.lang.Object

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

Field Detail

EQUAL_RANGE

public static String EQUAL_RANGE

EQUAL_NUMBER

public static String EQUAL_NUMBER

MEAN_STDEV

public static String MEAN_STDEV

MAX_BREAKS

public static String MAX_BREAKS

JENKS_BREAKS

public static String JENKS_BREAKS

KMEANS_OPTIMIZE

public static String KMEANS_OPTIMIZE

Constructor Detail

Classifier1D

public Classifier1D()

Method Detail

getAvailableClassificationMethods

public static List getAvailableClassificationMethods()

classifyEqualRange

public static double[] classifyEqualRange(double[] data,
                                          int numberClasses)

calculates class limits with equal range

Parameters:

data - input data

numberClasses - number of classes

Returns:

break values for classes. E.g. for 4 ranges 3 breaks are returned. Min and Max Values are not returned.

classifyEqualNumber

public static double[] classifyEqualNumber(double[] data,
                                           int numberClasses)

calculates class limits with equal number, which is euqal to the "quantiles" method. Note that differences in the items per classes occure, if items have same values and need to be grouped into the same class.

Parameters:

data - input data

numberClasses - number of classes

Returns:

break values for classes. E.g. for 4 ranges 3 breaks are returned. Min and Max Values are not returned.

classifyMeanStandardDeviation

public static double[] classifyMeanStandardDeviation(double[] data,
                                                     int numberClasses)

calculates class limits using mean value and standard deviation, i.e. for 5 classes: c1: values < m- 2std, c2: m - 2std < values < m - 1std, c3: m - 1std < values < m + 1std, c4: m + 1std < values < m + 2std c5: values > m- 2std

Parameters:

data - input data

numberClasses - number of classes

Returns:

break values for classes. E.g. for 4 ranges 3 breaks are returned. Min and Max Values are not returned.

classifyMaxBreaks

public static double[] classifyMaxBreaks(double[] data,
                                         int numberClasses)

calculates class limits using Maximum Breaks method (see e.g. T. A. Slocum: "Thematic Cartography and Visualization", 1999)

Parameters:

data - input data

numberClasses - number of classes

Returns:

break values for classes. E.g. for 4 ranges 3 breaks are returned. Min and Max Values are not returned.

classifyNaturalBreaks

public static double[] classifyNaturalBreaks(double[] data,
                                             int numberClasses)

calculates class limits using Jenks's Optimisation Method(Natural Break)

Parameters:

data - input data

numberClasses - number of classes

Returns:

break values for classes. E.g. for 4 ranges 3 breaks are returned. Min and Max Values are not returned.

classifyKMeansOnExistingBreaks

public static double[] classifyKMeansOnExistingBreaks(double[] data,
                                                      int numberClasses,
                                                      int initialLimitAlgorithm)

calculates class limits using optimal breaks method (see e.g. T. A. Slocum: "Thematic Cartography and Visualization", 1999, p.73) or B.D. Dent: "Cartography: Thematic Map Design", 1999, p.146).

Note: limits should not be equal to values. Since values that are equal to bounds can be classified into 2 classes.

Parameters:

data - input data

numberClasses - number of classes

initialLimitAlgorithm - 1: maxBreaks, 2: equalRange, 3: quantiles, 4: MeanStd-Dev 5: Jenks

Returns:

break values for classes. E.g. for 4 ranges 3 breaks are returned. Min and Max Values are not returned.

adjustLimitsKMeans

public static double[] adjustLimitsKMeans(double[] data,
                                          double[] oldLimits)

Moves the limits, by assigning data points to the closest class mean value.

This approach is equal to the k-means procedure (see e.g. Duda, Hart and Stork 2000, p. 526).

Parameters:

data - (sortedData from min to max, e.g. use jmathtools DoubleArray.sort())

oldLimits - old limits array

Returns:

a double array of adjusted limits

classifyData

public static int[] classifyData(double[] data,
                                 double[] limits)

Classifies the given data according to the given limits.

Parameters:

data - input data

limits - The break/decision values between the classes. Highest and lowest values are not delivered. Example Limits are for instance delivered by the Classifier1D.classifyEqualNumber() method.

Returns:

array containg a class ID for every item.

isInClass

public static boolean isInClass(double val,
                                double lowerBound,
                                double upperBound)

Checks if value is within limits.

Note: values equal to the bound values return "true". (qery: lowerlimit <= val <= upperlimit)

Parameters:

val - the value to test

lowerBound - the lower bound

upperBound - the upper bound

Returns:

true if val is included between lowerBound (included) and upperBound (included)

calcSDAM

public static double calcSDAM(double[] data)

SDAM (squared deviation [from] array mean): see B.D. Dent (1999, p. 148) alternatively look for T.A. Slocum (1999, p. 73).

Used for Optimal Breaks Method.

Parameters:

data - input data

Returns:

the squared deviation from double array mean

calcSDCM

public static double calcSDCM(double[] data,
                              int[] classes,
                              double[] classMeans,
                              int numClasses)

SDCM (squared deviations [from] class means): see B.D. Dent (1999, p. 148) alternatively look for T.A. Slocum (1999, p. 73). \n Used for Optimal Breaks Method. TODO : definition of SDCM (relative to SDAM)

Parameters:

data - input data

classes - the classes for every item of the data array

classMeans - class means

numClasses - number of classes

Returns:

squared deviations from class means

calcGVF

public static double calcGVF(double SDAM,
                             double SDCM)

GVF (goodness of variance fit): see B.D. Dent (1999, p. 148) alternatively look for T.A. Slocum (1999, p. 73). \n Used for Optimal Breaks Method.

Parameters:

SDAM - squared deviation [from] array mean

SDCM - squared deviation [from] class mean

Returns:

the Goodness of Variant Fit for a particular SDAM and SDCM

calcGVF

public static double calcGVF(double[] data,
                             double[] limits,
                             double SDAM)

GVF (goodness of variance fit): see B.D. Dent (1999, p. 148) alternatively look for T.A. Slocum (1999, p. 73). \n Used for Optimal Breaks Method.

Parameters:

data - input data

limits - The break/decision values between the classes. Highest and lowest values are not delivered. Example Limits are for instance delivered by the Classifier1D.classifyEqualNumber() method.

SDAM - squared deviation [from] array mean

Returns:

goodness of variance fit

calcClassMeans

public static double[] calcClassMeans(double[] data,
                                      int[] classes,
                                      int numClasses)

Parameters:

data - input data

classes - the vector containing the information on the class for an item

numClasses - the number of classes

Returns:

class means