This is a Python implementation of the Optigrid algorithm described in "Optimal Grid-Clustering: Towards Breaking the Curse of Dimensionality in High-Dimensional Clustering" published by Hinneburg and Keim in: "Proceedings of the 25th International Conference on Very Large Databases", 1999, pp. 506-517

This implementation is limited to projections to the coordinate axes and uses the scikit-learn Kernel Density estimation to determine the best cutting planes. In contrary to the article we use a maximum cut score instead of a minimum one, because the score is just the density at the cutting point, and we want to keep that as low as possible.

An example can be found in the demo.py file. To use the algorithm in your project, just copy the files optigrid.py and grid_level.py into your working tree.