I've noticed the Levenberg-Marquardt algorithm is only used with least squares problem and I didn't find any library in R or Python which allow to minimise the absolute values of the residual (and not the square of the residual)

  1. Is there any reason of that ?
  2. Does it make sense to apply sqrt(abs(residual)) on the residual before it's been squared if we want to use this algorithm with Least Absolute Deviation instead of Least Squares Deviation ?
  • $\begingroup$ Is it because sqrt(abs(x)) is not differentiable when x=0 ? $\endgroup$ – psql Oct 28 '15 at 11:48

Because it is based on a Second Order Approximation of the Squared Residual Function.

Hence it requires "Squared Residual Function".
The method will fit any model you can make which has the same form.

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.