levenberg-marquardt in lsqcurvefit

Hello

How do I use the levenberg-marquandt algortime in lsqcurve fit instead of the default trust-region-reflective algorithm?

[parameters,resnorm,residual,exitflag,output] = lsqcurvefit(heightModel,initialValues,filteredData(:,1),filteredData(:,2));

is the formula I use in my function

Can anyone help me out?

Thanks in advance

Niels

Answer 1

[Matt Kindig]

Pass in an options structure:

opts = optimset('Algorithm', 'levenberg-marquardt');
[parameters,resnorm,residual,exitflag,output] = lsqcurvefit(heightModel,initialValues,filteredData(:,1),filteredData(:,2),[],[],opts);

You can see the various options in the Help docs for lsqcurvefit, at

doc lsqcurvefit

Answer 2

[Niels]

Thanks for answering

I have used this for individual fitting in the following construct:

CODE PART A

    validFitPersons = true(nbValidPersons,1);
      for i=1:nbValidPersons
          personalData = data{validPersons(i),3};
          personalData = personalData(personalData(:,1)>=minAge,:);
          % Fit a specific model for all valid persons
          try
              opts = optimset('Algorithm', 'levenberg-marquardt');
             [personalParams,personalRes,personalResidual] = lsqcurvefit(heightModel,initialValues,personalData(:,1),personalData(:,2),[],[],opts);

          %catch
              %x=1;
          end

The intial starting values and the formula used to fit is given by:

CODE PART B

strcmpi(model,'jpa2')
    % y = a.*(1-1/(1+(b_1(t+e))^c_1+(b_2(t+e))^c_2+(b_3(t+e))^c_3))
    heightModel = @(params,ages) abs(params(1).*(1-1./(1+(params(2).* (ages+params(8) )).^params(5) +(params(3).* (ages+params(8) )).^params(6) +(params(4) .*(ages+params(8) )).^params(7) )));
    modelStrings = {'a','b1','b2','b3','c1','c2','c3','e'};

      % Define initial values
      if strcmpi('male',gender)
          initialValues = [175 0.35 0.12 0.076 0.43 2.8 20 0.34];
      else
          initialValues = [162 0.43 0.13 0.089 0.48 2.9 16 0.4];
      end

Switching off the catch command in CODE PART A and adding the levenberg-marquandt command also in CODE PART A caused my data to be much closer towards the initial values then otherwise. This is a good thing.

How can I make my curve fitting stay looking much closer around my initial values and not spreading out too much?

Anyhelp?