Closed lindahua closed 10 years ago
I'm alright with that. We put it in here originally because we thought we would use the characteristic function of distributions to allow different kernels. Since we hardcoded a Gaussian kernel, it seems reasonable to move this to Stats.jl.
Implemented by b4ecede
Thanks for doing this!
I am in the process of doing some refactoring of the Stats.jl package.
kernel density estimation is mostly data statistics. It does not directly related to a parametric distribution.
Therefore, I think Stats.jl might be a more appropriate place for it. (Like the
ecdf
function that is in Stats).cc: @johnmyleswhite