Closed wildart closed 10 years ago
Fitting seems a bit too ambitious. At some point we'll want to have a larger package that merges tools for machine learning. For now, I'm just thinking of this as a package to store methods that depend only on distances.
It is not limited only to kNN algorithms, e.g. kernel regression. Maybe give it more general name, such as "Fitting" or its derivatives, by analogy with Matlab Curve Fitting Toolbox, and include various models for regression, interpolation, smoothing, etc.