confidence intervals with lsqlin

Hi all,

I'd like to know if a function exists that computes the intervals of confidence of the result of the lsqlin optimisation given the intervals of confidence of the input data.

Using the notations here (<http://www.mathworks.fr/fr/help/optim/ug/lsqlin.html)>, I'm basically looking for confidence intervals of x components given the confidence intervals of d components.

Thanks in advance for your advice!

Answer 1

[Shashank Prasanna]

If you have constraints (looks like you do since you use lsqlin) then you will have to explore bootstrapping methods to estimate confidence intervals in the presence of constraints.

http://www.mathworks.com/help/stats/bootci.html

If you simply have bound constraints then you can transform your data as explained here:

http://www.mathworks.com/matlabcentral/answers/102066-how-do-i-find-the-confidence-intervals-for-the-estimated-coefficients-of-a-bounded-nonlinear-regress

If you do not have constraints or bounds, simply use fitlm which give you all the information you may need to assess your fit:

http://www.mathworks.com/help/stats/fitlm.html

[Jean-Yves]

Thanks, as you suspected, I'm in the first case, so I'll look into bootci!

[Hermence Dewintre]

Have you found how to have the confidence interval ? I'm in the same situation as you and i can't understand how bootci works

[Rohan Kulkarni]

Same here! could someone please explai how bootci could be applied to our cases?

[chen zhihua]

fx=@(x)lsqlin(C,d,A,b,Aeq,beq,lb,ub,options);

ci=bootci(nboot,fx,C,d,...)