Genesis of the Yu'erya gold deposit in eastern Hebei: Constraints from machine learning evidence in pyrite geochemistry

Yu'erya gold deposit is a representative one in the eastern Hebei region. Previous studies have investigated the source of the ore-forming materials, the age of mineralization and the fluid inclusions of the deposit. However, there is still controversy over its genesis. Combined with the random forest and support vector machine algorithms focusing on pyrite (4544 sets of data) from magmatic hydrothermal and orogenic gold deposits worldwide, this study conducted in-situ analysis of the compositions of the gold-related pyrite using laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) to reveal its genetic type. The LA-ICP-MS analysis results show that the pyrite in Yu'erya is characterized by enrichment of Co, Au, Ag, Cu, As, Pb and Te, which is consistent with that of typical magmatic hydrothermal gold deposits. Traditional discriminant diagrams based on Co-Ni and Ni-Co-As ratios exhibit ambiguity and are ineffective in distinguishing pyrite genetic types. The machine learning model constructed in this study achieved a classification accuracy of 93% through ten-fold cross-validation, significantly outperforming the traditional method. Combined with the existing mineralogical and chronological evidence in the mining area and the prediction results, it is confirmed that the Yu'erya gold deposit is of typical magmatic hydrothermal origin. This study verified the effectiveness of machine learning methods in integrating high-dimensional pyrite geochemical data, discriminating ore deposit genesis and reconstructing the mineralization process, providing new constraints for the analysis of complex mineralization systems. It further deepened the understanding of the genesis of the Yu'erya gold deposit which is critical for supporting the exploration of regional gold deposits.