Serial and parallel bindings to brute force and kd-tree methods for low dimensional (<16) nearest neighbor problems in python
Some basic benchmarks have been carried out against sklearn.neighbors to show the relative performance
of this library. These preliminary results show better performance and better scaling in every function.
However, keep in mind that sklearn handles a generic n-dimensional space while this package
has been simplified to work with 2D data. Moreover, sklearn also has an additional algorithm ball_tree that
scales to higher dimensions (N > 15) much better than kd-trees.
