On 11/14/2011 04:00 AM, Nir Krakauer wrote: > I've written a function (attached) to implement a nonparametric > (kernel-based) monotone increasing regression method. Please have a > look and add it to the Statistics package if it seems appropriate.
Thank you very much! This is a very nice contribution. I would commit it as it is. But if you would like to spend more time on the function, here are a couple of comments/suggestions: (1) It would be nice if you would say in the function description that you use a Epanechnikov kernel. (2) You assume that x and y are vectors. This could be checked. (3) In the inner loop, 150 is the maximum number of iterations. Maybe I am wrong but I think this number should not be reached on success. Should there be a warning if the loop is not terminated by the break statement? Is the +0.1/-0.1 condition robust for different scales of input data? Would it make sense to have the maximum number of iterations as an optional function argument? Thanks, Arno ------------------------------------------------------------------------------ RSA(R) Conference 2012 Save $700 by Nov 18 Register now http://p.sf.net/sfu/rsa-sfdev2dev1 _______________________________________________ Octave-dev mailing list Octave-dev@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/octave-dev
