I'm skeptical about the general usefulness of the tool, but I feel like the best way to evaluate such suspicions to to find a fairly well-understood dataset and compare the utility of traditional measures of correlations vs MIC in finding true relationships in the data.

MINE: Free tool for detecting novel associations in large data sets

by curiousepic 1 min read17th Dec 20115 comments


I was waiting for someone more knowledgeable to post something about this, but it's been a couple days and thought I'd bring it to LW's attention.

The maximal information coefficient (MIC) is a measure of two-variable dependence developed with the guidelines of generality and equitability in mind. The published paper describing MIC shows that it comes very close to achieving both goals simultaneously, and that it significantly outperforms competing methods in this regard.

Video summary

Main site with access to tool