Application of Machine Learning to Hydrothermal System Analysis: Geochemical Insights from the Bektakari–Bneli Khevi Ore Knot, Southern Georgia

This study integrates geochemical, statistical, and machine learning methods to investigate hydrothermal systems and mineralization processes within southern Georgia's Bektakari–Bnelikhevi ore knot. A total of 212 geochemical samples were analyzed, revealing key elemental associations such as V–Sc, Mo–W, and S–V, indicative of magmatic-hydrothermal activity and metasomatic alteration, including albitization and potassic enrichment. Principal Component Analysis (PCA) and DBSCAN clustering identified two dominant alteration regimes: sulfide-rich mineralization and alkali metasomatism. Geochemical indices, Alteration Index (AI) and Chlorite–Carbonate–Pyrite Index (CCPI), effectively delineate alteration zones. AI values ranged from 45 to 95, while CCPI ranged from 30 to 85, with the highest mineralization potential concentrated in sericitic and Na–Ca zones. Feature importance analysis highlighted the Cu–Ag–Pb Index (32%) and Metallicity Factor (27%) as the strongest predictors of mineralized zones. Machine learning models achieved high precision in identifying epithermal and porphyry zones (Precision 0.85), though recall remained low in transitional areas (Recall ~0.38), suggesting underrepresentation or overlapping features in these zones. This integrated approach offers a data-driven framework for targeting hydrothermal mineralization. The findings can inform exploration strategies by prioritizing geochemical signatures and improving zone classification in complex alteration systems.