nonlinear curve-fitting: weight vector

hi I am working on a multiobjective procedure using the 'lsqcurvefit' (Levenberg-Marquardt algorithm): I would like to use a weight vector during risolution of the best fit in order to force the best fit function following the last inputs in a better way instead of the first ones. How can I?

Answer 1

[Richard Willey]

nlinfit in Statistics Toolbox

1. Uses Levenberg-Marquardt under the hood
2. Allows you to input a vector of weights

See http://www.mathworks.com/help/toolbox/stats/nlinfit.html

[Tom Lane]

The use of weights is actually not supported directly in nlinfit, but there is a demo about how to do that. You might use the same technique with lsqcurvefit if you prefer.

https://www.mathworks.com/help/stats/examples/weighted-nonlinear-regression.html

[anna]

Dear guys,

many many thanks for your reply.

It works perfectly!!!

I have modified my script by using the same technique reported in the last link for the “nlinfit”.