https://youtu.be/ArdPDHZqfOc

Mineral exploration seeks to uncover valuable deposits, which often have distinctive element compositions compared to their surrounding geological environment. Identifying buried or sub-outcropping deposits, particularly in areas with transported cover, poses a significant challenge. Local outlier detection methods offer a powerful approach by pinpointing samples that are anomalous relative to their spatial neighbors, potentially highlighting zones of mineralization. Effectively applying these methods to geochemical data requires recognizing its fundamental compositional nature, where the crucial information lies in the relative proportions and ratios of elements, not just absolute values. Specialized techniques from compositional data analysis (CoDA), such as logratio transformations, are essential preprocessing steps to ensure valid statistical analysis of this type of data. This study investigates the performance of several local outlier detection methods using real-world geochemical compositional data sets that vary in scale, sampling density, and data quality. It underscores the critical role of careful data preprocessing and the appropriate application of CoDA techniques. The research evaluates how well different methods identify known mineralizations and discusses the influence of factors like data quality, the suite of elements analyzed, and sampling density. A key finding is that data quality and the availability of a sufficient set of elements are more important for identifying mineral deposits than high sampling density alone. Ultimately, statistical analysis using these methods is most effective when combined with diagnostic tools and expert geological interpretation to validate promising exploration targets. P. Geo. Ricardo A Valls, M. Sc. and Geo Gadfly Valls Geoconsultant ORCID ID- https://orcid.org/0000-0002-5421-0914 Scopus Author ID: 7003369619/35335510700 ResearcherID: S-6604-2018 If you like this content, please "buy me a coffee" https://www.buymeacoffee.com/goldendroplets