Orthogonal Linear Regression in 3D-space by using Principal Components Analysis
This is a wrapper function to some pieces of the code from the Statistics Toolbox demo titled "Fitting an Orthogonal Regression Using Principal Components Analysis"
(http://www.mathworks.com/products/statistics/demos.html?file=/products/demos/shipping/stats/orthoregdemo.html),
which is Copyright by the MathWorks, Inc.
Input parameters:
- XData: input data block -- x: axis
- YData: input data block -- y: axis
- ZData: input data block -- z: axis
- geometry: type of approximation ('line','plane')
- visualization: figure ('on','off') -- default is 'on'
- sod: show orthogonal distances ('on','off') -- default is 'on'
Return parameters:
- Err: error of approximation - sum of orthogonal distances
- N: normal vector for plane, direction vector for line
- P: point on plane or line in 3D space
Example:
>> XD = [4.8 6.7 6.2 6.2 4.1 1.9 2.0]';
>> YD = [13.4 9.9 5.8 6.1 6.7 10.6 11.5]';
>> ZD = [13.7 13.1 11.3 11.8 12.5 16.2 18.5]';
>> fit_3D_data(XD,YD,ZD,'line','on','on');
>> fit_3D_data(XD,YD,ZD,'plane','on','off');
Note: Written for Matlab 7.0 (R14) with Statistics Toolbox
We sincerely thank Peter Perkins, the author of the demo, and John D'Errico for their comments.
Ivo Petras, Igor Podlubny, May 2006
(ivo.petras@tuke.sk, igor.podlubny@tuke.sk)
An example of application can be found at:
http://uk.arxiv.org/abs/math/0609789
For additional codes for the Orthogonal Linear Regression also known as Total Least Squares Method see link:
http://www.mathworks.com/matlabcentral/fileexchange/31109 |
