CDMatrixStat

Class Description

The CDMatrixStat class provides functions calculating statistics for the matrix and some statistical tests.

• Constructors/Destructor
• Methods
 
• Matrix Statistics
 

Mean	MeanCol	Std
StdCol	Sum	SumCol
Var	VarCol


 
• Statistical Tests

OneWayANOVA

TTest


 
• Miscellaneous

ToString

• Overloaded Operators

CDMatrixStat (void)

Default constructor. Creates an uninitialized matrix.

CDMatrixStat
    (int rows, int cols,
     double *storage = 0)

If storage is zero, create a dynamically sized double precision matrix with the number of rows and cols specified. If storage is not zero, use the memory it points to.

CDMatrixStat (const CDMatrixStat &mat)

Creates a matrix with the same properties and values as another matrix.
 

Argument	Description
rows,cols	number of rows and columns in matrix
storage	pointer to previously allocated memory
mat	reference to another matrix

~CDMatrixStat (void)
Destructor.
 

Top

Matrix Statistics

CDMatrixStat::Mean

double Mean()

Return the mean value of the elements.

Top

void MeanCol(CDVectorStat& vMean)

Calculate the mean value of the elements in each column.
 

Argument	Description
vMean	reference to vector holding the mean values

Top

CDMatrixStat::Std

double Std()

Return the standard deviation of the elements.

Top

void StdCol(CDVectorStat& vStd)

Calculate the standard deviation of the elements in each column.
 

Argument	Description
vStd	reference to vector holding the standard deviations

Top

double Sum()

Return the sum of the elements.

Top

void SumCol(CDVectorStat& vSum)

Calculate the sum of the elements in each column.
 

Argument	Description
vSum	reference to vector holding the sums

Top

double Var()

Return the variance of the elements.

Top

void VarCol(CDVectorStat& vVar)

Calculate the variance of the elements in each column.
 

Argument	Description
vVar	reference to vector holding the variance

Top

Statistical Tests

CDMatrixStat:: OneWayANOVA

void OneWayANOVA(double& dZ, int& nDFModel, int& nDFError)

Perform a one way analysis of variance, where each column in the matrix corresponds to a set of observations, which are assumed to be Gaussian distributed, i.e.

Hypothesis:    mean1 = mean2 = mean3 = ...

Reject region:    dz > F(nDFModel, nDFError )_{1-alpha}
 

Argument	Description
dZ	test statistics
nDFModel	degrees of freedom for the model
nDFError	degrees of freedom for the error

 

void TTest(const int iCol1,const int iCol2, double& dZ, int& nDF)

Perform a t-test on the observations corresponding to columns iCol1 and iCol2, which are assumed to be Gaussian distributed observation,  i.e.

Hypothesis:    mean1 = mean2

Reject region:    dZ < t(nDF)_alpha    v    dZ > t(nDF)_{1-alpha}

Note, that the variance estimate used in the t-test is created by a pulled variance based on all the columns in the matrix.
 

Argument	Description
iCol1, iCol2	index to columns, which corresponds to the two set of observations
dZ	test statistics
nDF	degrees of freedom

Miscellaneous

CDMatrixStat::ToString

void ToString(CString& strOut)

This function returns a string representing the vector.
 

Argument	Description
strOut	reference to the string

CDMatrixStat::operator=

CDMatrixStat& operator= (double value)
Assigns value to all elements of this matrix.

CDMatrixStat& operator= (const CDMatrixStat& vec)
Copies the values of another matrix to the corresponding elements of this matrix.
 

Argument	Description
value	value to be assigned to this matrix's elements
vec	reference to matrix with values to be copied

Copyright, 1999
Section for Image Analysis
Department of Mathematical Modelling
Technical University of Denmark
DK-2800 Lyngby

Authors: Rune Fisker

Last updated: 03-05-1999