is there something similar to Excel Solver in Matlab?

Hi, I have a similar problem. To simplify assume a linear equation y=mx+c. I have values for the independent variable x, the actual y and want to solve for coefficients m and c. Using Excel Solver I would use random initial values for m and c in my equation, and get a fitted y, say y_fit. Then work the sum of squared residuals between y and y_fit [RSS = sum((y-yfit)^2)]. Then tell Solver to give me a solution for m and c which minimises RSS. Is this possible in Matlab? Can I make use of the Optimisation tool box to do this?

Answer 1

[Titus Edelhofer]

Hi,

take a look at lsqcurvefit http://www.mathworks.com/help/toolbox/optim/ug/lsqcurvefit.html from Optimization Toolbox. It should do what you are looking for ...

Titus

[Andrew]

Thanks Titus. That's what I was looking for. It worked:)

Andrew

[John C L Mayson II]

Hi Andrew, I am a beginner in MATLAB and currently struggling with the same problem. I visited the link uploaded by Titus but still couldn't figure it out. I have a set of equations which obtains a variable "b". In one of those equations, I assumed a constant value for a variable "a". Now I want to set variable "b" to 0.01 by changing variable "a". Can you please help if you have the time?

Answer 2

[Fangjun Jiang]

From help robustfit.

       x = (1:10)';
       y = 10 - 2*x + randn(10,1); y(10) = 0;
       bls = regress(y,[ones(10,1) x])
       brob = robustfit(x,y)
       scatter(x,y)
       hold on
       plot(x,brob(1)+brob(2)*x,'r-', x,bls(1)+bls(2)*x,'m:')

[Andrew]

Thanks Fangjun. Problem is that I want to apply solution to a wider problem that may not necessarily be minimising sum of squared residuals of a linear equation. So in this case, is it possible to minimise RSS by changing m and c by means of some iterative process?

Thanks once again.

[Fangjun Jiang]

I am not following. If you assume y=m*x+c, it means linear and the result from regress() is the result of minimizing RSS. If you want to do for example, y=n*x^2+m*x+c, then you can use regress(y,[ones(10,1) x x.^2]). There is no need to do iteration.