This project aims to be an open source implementation of the maximal information coefficient measure, a measure of two-variable dependence designed for the exploration of many-dimensional data sets. The project's source code is distributed under the permissive Apache License.
For more information on the measure see the web site created by the team that came up with the measure and the corresponding article they published in Science.
Currently the code implements all the algorithms listed in the article, compiles, and passes all unit tests.
Things that remain to be done are:
- Correct the implementation to produce results close to those of the original implementation
- Optimize the algorithm through aggressive caching
- Explore other methods to obtain better results in less time