Saw a talk about using more "big data" means in paleo. This group of scientists gave a talk about using affinity analyses to make structural measurements of a bunch of fish, amphibian and reptile fossils and theirliving forms. the affinities qere based on the number of similar traits shared by different species (Unfortunately theres no genetic data from fossils so it hd to be presented very tentatively.
But the affinity analyses showed that, when the computer is allowed to arrange the affinities the derived line is Time and show the screwey trends that evolution took. The Story of Earth is a new book that an author from Carnegie Tech wrote an explained the use of the technique an how its being used for lots of area in science (like the order in which mineral appeared on the planet ). Im thinkingof using the technique to help determine , from mineral affinities, where would we expect to find certain minerls.
Theres only about 5500 mineral species on earth right now and millions of affinity trends that can be mined.
Did your work in puter science mess with Affinity analyses, specifically "market basket".
I cme close in Krigging or variogramming but this trick allows the linkage of many more parameters into "affinity webs"