Mineral Association Analysis: Predicting unknown mineral occurrences and improving our understanding of mineralogy

Abstract

Minerals are the oldest surviving materials from the formation of our solar system. They are time capsules that store and provide information about the evolution of Earth and other planetary bodies. In addition to being a cornerstone of geoscience research, minerals also have economic, industrial and commercial importance in many sectors of society. One of the fundamental questions in mineralogy and geosciences in general is “Where to find minerals?”. Due to the complex and intertwined nature of natural systems, it has been hard to predict the occurrences of minerals. However, with increase in the volume and accuracy of mineral data and rise of mineral informatics, data science and analytics methods can be developed to answer this fundamental question in mineralogy.  In this contribution, we present “mineral association analysis”, a method to: 1) Predict the mineral inventory for any existing locality. 2) Predict previous unknown localities for any given mineral. Mineral association analysis is a machine learning method that uses association rule learning to find interesting patterns based on mineral occurrence data. Using mineral association analysis, we have been able to predict locations of critical minerals, such as minerals with Li- and Th-bearing phases, predict the mineral inventory of mars analogue sites, and even understand how mineralization and mineral associations changed through deep time.