Interpretation of Trace Element Chemistry of Zircons from Bor and Cukaru Peki: Conventional Approach and Random Forest Classification

Abstract

The deposits of Bor and Cukaru Peki are important contributors to the Apuseni–Banat–Timok–Srednogorie (ABTS) belt’s metallogenic endowment. We use decision tree and random forest algorithms applied to zircon geochemistry data from Bor, Cukaru Peki and a selection of other localities within the ABTS. The resulting predictions, supported by high scores on the test set predictions for the random forest algorithm, suggest that it is possible to fingerprint the studied deposits and localities from the ABTS belt based on zircon geochemistry. These results take into account the multivariate geochemical patterns and can be used in combination with a widely accepted Eu anomaly indicator or assist in finding more subtle geochemical differences for systems where applying a single cut-off value does not result in a good separation between barren and mineralized rocks.

Keywords

• zircon geochemistry
• machine learning
• Timok magmatic complex
• Bor
• Cukaru Peki