Package weka.classifiers

Class CostMatrix

java.lang.Object
weka.classifiers.CostMatrix
All Implemented Interfaces:
Serializable, RevisionHandler
public class CostMatrix extends Object implements Serializable, RevisionHandler
Class for storing and manipulating a misclassification cost matrix. The element at position i,j in the matrix is the penalty for classifying an instance of class j as class i. Cost values can be fixed or computed on a per-instance basis (cost sensitive evaluation only) from the value of an attribute or a mathematical expression involving attribute(s).

Values in an instance are accessed in an expression by prefixing their index (starting at 1) with the character 'a'. E.g.

a1 ˆ 2 * a5 / log(a7 * 4.0)
Supported opperators: +, -, *, /, ^, log, abs, cos, exp, sqrt, floor, ceil, rint, tan, sin, (, ).
Version:
$Revision: 11868 $
Author:
Mark Hall, Richard Kirkby (rkirkby@cs.waikato.ac.nz)
See Also:

  • Field Summary

    Fields
    Modifier and Type
    Field
    Description
    static String
    FILE_EXTENSION
    The deafult file extension for cost matrix files

  • Constructor Summary

    Constructors
    Constructor
    Description
    CostMatrix(int numOfClasses)
    Creates a default cost matrix of a particular size.
    CostMatrix(Reader reader)
    Reads a matrix from a reader.
    CostMatrix(CostMatrix toCopy)
    Creates a cost matrix that is a copy of another.

  • Method Summary

    Modifier and Type
    Method
    Description
    Instances
    applyCostMatrix(Instances data, Random random)
    Applies the cost matrix to a set of instances.
    double[]
    expectedCosts(double[] classProbs)
    Calculates the expected misclassification cost for each possible class value, given class probability estimates.
    double[]
    expectedCosts(double[] classProbs, Instance inst)
    Calculates the expected misclassification cost for each possible class value, given class probability estimates.
    final Object
    getCell(int rowIndex, int columnIndex)
    Return the contents of a particular cell.
    final double
    getElement(int rowIndex, int columnIndex)
    Return the value of a cell as a double (for legacy code)
    final double
    getElement(int rowIndex, int columnIndex, Instance inst)
    Return the value of a cell as a double.
    double
    getMaxCost(int classVal)
    Gets the maximum cost for a particular class value.
    double
    getMaxCost(int classVal, Instance inst)
    Gets the maximum cost for a particular class value.
    String
    getRevision()
    Returns the revision string.
    void
    initialize()
    Initializes the matrix
    void
    normalize()
    Normalizes the matrix so that the diagonal contains zeros.
    int
    numColumns()
    Same as size
    int
    numRows()
    Same as size
    static CostMatrix
    parseMatlab(String matlab)
    creates a matrix from the given Matlab string.
    void
    readOldFormat(Reader reader)
    Loads a cost matrix in the old format from a reader.
    final void
    setCell(int rowIndex, int columnIndex, Object value)
    Set the value of a particular cell in the matrix
    final void
    setElement(int rowIndex, int columnIndex, double value)
    Set the value of a cell as a double
    int
    size()
    The number of rows (and columns)
    String
    toMatlab()
    converts the Matrix into a single line Matlab string: matrix is enclosed by parentheses, rows are separated by semicolon and single cells by blanks, e.g., [1 2; 3 4].
    String
    toString()
    Converts a matrix to a string.
    void
    write(Writer w)
    Writes out a matrix.

    Methods inherited from class java.lang.Object

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

  • Field Details

    • FILE_EXTENSION

      public static String FILE_EXTENSION
      The deafult file extension for cost matrix files

  • Constructor Details

    • CostMatrix

      public CostMatrix(int numOfClasses)
      Creates a default cost matrix of a particular size. All diagonal values will be 0 and all non-diagonal values 1.
      Parameters:
      numOfClasses - the number of classes that the cost matrix holds.

    • CostMatrix

      public CostMatrix(CostMatrix toCopy)
      Creates a cost matrix that is a copy of another.
      Parameters:
      toCopy - the matrix to copy.

    • CostMatrix

      public CostMatrix(Reader reader) throws Exception
      Reads a matrix from a reader. The first line in the file should contain the number of rows and columns. Subsequent lines contain elements of the matrix. (FracPete: taken from old weka.core.Matrix class)
      Parameters:
      reader - the reader containing the matrix
      Throws:
      Exception - if an error occurs
      See Also:

  • Method Details

    • initialize

      public void initialize()
      Initializes the matrix

    • size

      public int size()
      The number of rows (and columns)
      Returns:
      the size of the matrix

    • numColumns

      public int numColumns()
      Same as size
      Returns:
      the number of columns

    • numRows

      public int numRows()
      Same as size
      Returns:
      the number of rows

    • applyCostMatrix

      public Instances applyCostMatrix(Instances data, Random random) throws Exception
      Applies the cost matrix to a set of instances. If a random number generator is supplied the instances will be resampled, otherwise they will be rewighted. Adapted from code once sitting in Instances.java
      Parameters:
      data - the instances to reweight.
      random - a random number generator for resampling, if null then instances are rewighted.
      Returns:
      a new dataset reflecting the cost of misclassification.
      Throws:
      Exception - if the data has no class or the matrix in inappropriate.

    • expectedCosts

      public double[] expectedCosts(double[] classProbs) throws Exception
      Calculates the expected misclassification cost for each possible class value, given class probability estimates.
      Parameters:
      classProbs - the class probability estimates.
      Returns:
      the expected costs.
      Throws:
      Exception - if the wrong number of class probabilities is supplied.

    • expectedCosts

      public double[] expectedCosts(double[] classProbs, Instance inst) throws Exception
      Calculates the expected misclassification cost for each possible class value, given class probability estimates.
      Parameters:
      classProbs - the class probability estimates.
      inst - the current instance for which the class probabilites apply. Is used for computing any non-fixed cost values.
      Returns:
      the expected costs.
      Throws:
      Exception - if something goes wrong

    • getMaxCost

      public double getMaxCost(int classVal) throws Exception
      Gets the maximum cost for a particular class value.
      Parameters:
      classVal - the class value.
      Returns:
      the maximum cost.
      Throws:
      Exception - if cost matrix contains non-fixed costs

    • getMaxCost

      public double getMaxCost(int classVal, Instance inst) throws Exception
      Gets the maximum cost for a particular class value.
      Parameters:
      classVal - the class value.
      Returns:
      the maximum cost.
      Throws:
      Exception - if cost matrix contains non-fixed costs

    • normalize

      public void normalize()
      Normalizes the matrix so that the diagonal contains zeros.

    • readOldFormat

      public void readOldFormat(Reader reader) throws Exception
      Loads a cost matrix in the old format from a reader. Adapted from code once sitting in Instances.java
      Parameters:
      reader - the reader to get the values from.
      Throws:
      Exception - if the matrix cannot be read correctly.

    • write

      public void write(Writer w) throws Exception
      Writes out a matrix. The format can be read via the CostMatrix(Reader) constructor. (FracPete: taken from old weka.core.Matrix class)
      Parameters:
      w - the output Writer
      Throws:
      Exception - if an error occurs

    • toMatlab

      public String toMatlab()
      converts the Matrix into a single line Matlab string: matrix is enclosed by parentheses, rows are separated by semicolon and single cells by blanks, e.g., [1 2; 3 4].
      Returns:
      the matrix in Matlab single line format

    • parseMatlab

      public static CostMatrix parseMatlab(String matlab) throws Exception
      creates a matrix from the given Matlab string.
      Parameters:
      matlab - the matrix in matlab format
      Returns:
      the matrix represented by the given string
      Throws:
      Exception
      See Also:

    • setCell

      public final void setCell(int rowIndex, int columnIndex, Object value)
      Set the value of a particular cell in the matrix
      Parameters:
      rowIndex - the row
      columnIndex - the column
      value - the value to set

    • getCell

      public final Object getCell(int rowIndex, int columnIndex)
      Return the contents of a particular cell. Note: this method returns the Object stored at a particular cell.
      Parameters:
      rowIndex - the row
      columnIndex - the column
      Returns:
      the value at the cell

    • getElement

      public final double getElement(int rowIndex, int columnIndex) throws Exception
      Return the value of a cell as a double (for legacy code)
      Parameters:
      rowIndex - the row
      columnIndex - the column
      Returns:
      the value at a particular cell as a double
      Throws:
      Exception - if the value is not a double

    • getElement

      public final double getElement(int rowIndex, int columnIndex, Instance inst) throws Exception
      Return the value of a cell as a double. Computes the value for non-fixed costs using the supplied Instance
      Parameters:
      rowIndex - the row
      columnIndex - the column
      Returns:
      the value from a particular cell
      Throws:
      Exception - if something goes wrong

    • setElement

      public final void setElement(int rowIndex, int columnIndex, double value)
      Set the value of a cell as a double
      Parameters:
      rowIndex - the row
      columnIndex - the column
      value - the value (double) to set

    • toString

      public String toString()
      Converts a matrix to a string. (FracPete: taken from old weka.core.Matrix class)
      Overrides:
      toString in class Object
      Returns:
      the converted string

    • getRevision

      public String getRevision()
      Returns the revision string.
      Specified by:
      getRevision in interface RevisionHandler
      Returns:
      the revision