How do i do a two dimensional linear regression fit

Hey,

I'm working with winddata and for a model i need to do a two-dimensional linear regression fit of the form

[y1;y2]=[a1;a2]+[b1,b2;b3,b4]*[x1;x2]

(x1,X2) and (y1,y2) are know and i want to determine to a and b coefficients.

Can anybody help me with this? The x en y coordinates are both 4392x2.

Thanks in advance!

Answer 1

[the cyclist]

Do you have the Statistics Toolbox? If so, I believe you can use the mvregress() function to do this.

[jdm_dm]

Hey,

Thanks for the quick response! But it doesn't seems to work. I've put the x and y coordinates in separate arrays respectively X and Y (both size 4392x2). When i do p=mvregress(Y,X)I get an error message: X must be a cell array if Y has multiple columns.

When I try to put X in a cell array (I call him mycell) and then try p=mvregress(Y,mycell) I get the following error: Undefined function 'isnan' for input arguments of type 'cell'.

Thanks,

Jdm

[the cyclist]

I suggest you look at the "flu" example here:

http://www.mathworks.com/help/toolbox/stats/mvregress.html

to help you debug your syntax.

Answer 2

[Image Analyst]

I use John D'Errico's polyfitn: http://www.mathworks.com/matlabcentral/fileexchange/?term=authorid%3A679

Answer 3

[Bahloul Derradji]

use the following ready to use example code:

this is the mainfile.m

clear all

clc

clf

close all

xdata =( -1:0.1:+1);

ydata=(-1:0.2:1)';

nx=numel(xdata);

ny=numel(ydata);

ax=0.8;

ay=0.4;

A=5;

[X,Y]= meshgrid(xdata,ydata);

Z=A*exp(-((X/ax).^2+(Y/ay).^2))+ 0.05*rand(ny,nx);

surf(X,Y,Z);

s = surf(X,Y,Z,'FaceAlpha',0.4);

s.EdgeColor = 'none';

s.FaceColor = 'red';

xlabel('x')

ylabel('y')

zlabel('z')

x = reshape(X,[],1);

y = reshape(Y,[],1);

z = reshape(Z,[],1);

%cftool

[fitresult, gof] = createFit(x, y, z);

disp(fitresult)

Here is the createFit.m script;

function [fitresult, gof] = createFit(x, y, z)

%% Fit: 'Gaussian fit 1'.

[xData, yData, zData] = prepareSurfaceData( x, y, z );

% Set up fittype and options.

ft = fittype( 'a*exp(-(x/wx)^2-(y/wy)^2)', 'independent', {'x', 'y'}, 'dependent', 'z' );

opts = fitoptions( 'Method', 'NonlinearLeastSquares' );

opts.Display = 'Off';

opts.StartPoint = [0.890036233228213 0.330202242514021 0.22970119787112];

% Fit model to data.

[fitresult, gof] = fit( [xData, yData], zData, ft, opts );

% Plot fit with data.

figure( 'Name', 'Gaussian fit 1' );

h = plot( fitresult, [xData, yData], zData );

legend( h, 'Gaussian fit 1', 'z vs. x, y', 'Location', 'NorthEast' );

% Label axes

xlabel x

ylabel y

zlabel z

grid on

% enjoy.