Qmin – A machine learning-based application for processing and analysis of mineral chemistry data

Abstract Mineral chemistry analysis is a valuable tool with many applications in mineralogy and mineral prospecting and beneficiation studies. This type of analysis can point out relevant information, such as the concentration of the chemical element of interest in the analyzed phase and, thus, the predisposition of an area for a given commodity. Therefore, considerable amounts of data have been generated, especially with the use of electron probe microanalyzers (EPMA), either for academic research or for prospecting and applied mineralogical work in the mineral industry. We have identified an efficiency gap when manually processing and analyzing mineral chemistry data, and thus, investigated the possibility that such research might benefit from the versatility of machine learning algorithms. We present Qmin, an application that assists in increasing the efficiency of processing and analysis of mineral chemistry data through automated routines. Our code benefits from a hierarchical structure of classifiers and regressors trained by a Random Forests algorithm developed on a filtered training database extracted from the GEOROC (Geochemistry of Rocks of the Oceans and Continents) repository, which is maintained by the Max Planck Institute for Chemistry. To test the robustness of our application, we applied a blind test with more than 22,000 mineral chemistry analyses compiled for diamond prospecting within the scope of the Diamante Brasil Project of the Geological Survey of Brazil. The blind test yielded a balanced classifier accuracy of ∼99% for the minerals known by Qmin. This outcome emphasizes the potential of machine learning techniques in assisting the processing and analysis of mineral chemistry data.