Linear regression analysis


Functions

template<typename T, int Size>
void computeFittingHyperplane (int numPoints, const Vector< T, Size > *points, Vector< T, Size+1 > *retCoefficients)
template<typename T>
void computeFittingHyperplaneX (int numPoints, const VectorX< T > *points, VectorX< T > *retCoefficients)
template<typename T, int Size>
void linearRegression (int numPoints, const Vector< T, Size > *points, Vector< T, Size > *retCoefficients, int funcOfOthers)
template<typename T>
void linearRegressionX (int numPoints, const VectorX< T > *points, VectorX< T > *retCoefficients, int funcOfOthers)

Detailed Description


Function Documentation

void Eigen::computeFittingHyperplane ( int  numPoints,
const Vector< T, Size > *  points,
Vector< T, Size+1 > *  retCoefficients 
) [inline]

This function is quite similar to linearRegression(), so we refer to the documentation of this function and only list here the differences.

The main difference from linearRegression() is that this function doesn't take a funcOfOthers argument. Instead, it finds a general equation of the form

\[ r_0 x_0 + \cdots + r_{n-1}x_{n-1} + r_n = 0, \]

where $n=Size$, $r_i=retCoefficients[i]$, and we denote by $x_0,\ldots,x_{n-1}$ the n coordinates in the n-dimensional space.

Thus, the vector retCoefficients has size $n+1$, which is another difference from linearRegression().

See also:
linearRegression()

void Eigen::computeFittingHyperplaneX ( int  numPoints,
const VectorX< T > *  points,
VectorX< T > *  retCoefficients 
) [inline]

This function is the dynamic-size counterpart to computeFittingHyperplane() and, aside from working with VectorX instead of Vector, is exactly the same thing.

See also:
linearRegressionX()

void Eigen::linearRegression ( int  numPoints,
const Vector< T, Size > *  points,
Vector< T, Size > *  retCoefficients,
int  funcOfOthers 
) [inline]

Performs a multiple linear regression on a set of points, as explained here:

http://en.wikipedia.org/wiki/Linear_regression#Multiple_linear_regression

In other words, for a set of points, this function tries to express one of the coords as a linear (affine) function of the other coords.

This is best explained by an example. This function works in full generality, for points in a space of arbitrary dimension, and also over the complex numbers, but for this example we will work in dimension 3 over the real numbers (doubles).

So let us work with the following set of 5 points given by their $(x,y,z)$ coordinates:

    Vector3d points[5];
    points[0] = Vector3d( 3.02, 6.89, -4.32 );
    points[1] = Vector3d( 2.01, 5.39, -3.79 );
    points[2] = Vector3d( 2.41, 6.01, -4.01 );
    points[3] = Vector3d( 2.09, 5.55, -3.86 );
    points[4] = Vector3d( 2.58, 6.32, -4.10 );
Suppose that we want to express the second coordinate ($y$) as a linear expression in $x$ and $z$, that is,

\[ y=ax+bz+c \]

for some constants $a,b,c$. Thus, we want to find the best possible constants $a,b,c$ so that the plane of equation $y=ax+bz+c$ fits best the five above points. To do that, call this function as follows:

    Vector3d coeffs; // will store the coefficients a, b, c
    linearRegression< double, 3 >( 5, points, & coeffs,
                                   1 // the coord to express as a function of
    // the other ones. 0 means x, 1 means y, 2 means z.
                                 );
Now the vector coeffs is approximately $( 0.495 , -1.927 , -2.906 )$. Thus, we get $a=0.495, b = -1.927, c = -2.906$. Let us check for instance how near points[0] is from the plane of equation $y=ax+bz+c$. Looking at the coords of points[0], we see that:

\[ax+bz+c = 0.495 * 3.02 + (-1.927) * (-4.32) + (-2.906) = 6.91.\]

On the other hand, we have $y=6.89$. We see that the values $6.91$ and $6.89$ are near, so points[0] is very near the plane of equation $y=ax+bz+c$.

Let's now describe precisely the parameters:

Parameters:
numPoints the number of points to read from the array
points the array of points on which to perform the linear regression
retCoefficients pointer to the vector in which to store the result. This vector must be of the same type and size as the data points. The meaning of its coords is as follows. For brevity, let $n=Size$, $r_i=retCoefficients[i]$, and $f=funcOfOthers$. Denote by $x_0,\ldots,x_{n-1}$ the n coordinates in the n-dimensional space. Then the result equation is:

\[ x_f = r_0 x_0 + \cdots + r_{f-1}x_{f-1} + r_{f+1}x_{f+1} + \cdots + r_{n-1}x_{n-1} + r_n. \]

funcOfOthers Determines which coord to express as a function of the others. Coords are numbered starting from 0, so that a value of 0 means $x$, 1 means $y$, 2 means $z$, ...
See also:
computeFittingHyperplane()

void Eigen::linearRegressionX ( int  numPoints,
const VectorX< T > *  points,
VectorX< T > *  retCoefficients,
int  funcOfOthers 
) [inline]

This function is the dynamic-size counterpart to linearRegression() and, aside from working with VectorX instead of Vector, is exactly the same thing.

See also:
computeFittingHyperplaneX()


Generated on Tue Mar 18 15:26:20 2008 for Eigen by  doxygen 1.5.5