Original scientific paper
https://doi.org/10.17794/rgn.2025.3.10
ADVANCED CLUSTERING TECHNIQUES FOR TIN DEPOSIT CLASSIFICATION IN MALAYSIA: A MACHINE LEARNING APPROACH
Meisam Saleki
; School of Materials and Mineral Resources Engineering, Engineering Campus, Universiti Sains Malaysia (USM), Malaysia
Kamran Mostafaei
orcid.org/0000-0002-4039-4628
; Department of Mining, Faculty of Engineering, University of Kurdistan, Sanandaj, Iran
*
Zakaria Bin Endut
; School of Materials and Mineral Resources Engineering, Engineering Campus, Universiti Sains Malaysia (USM), Malaysia
Mohammad Nabi Kianpour
; School of Materials and Mineral Resources Engineering, Engineering Campus, Universiti Sains Malaysia (USM), Malaysia
* Corresponding author.
Abstract
This study explores the application of advanced clustering techniques—Spectral Clustering, Gaussian Mixture Models (GMM), and a hybrid approach combining Autoencoders with K-Means—to classify tin deposits in Malaysia. Geochemical data from 28 tin samples across regions such as Pengkalan Hulu North, Menglembu, Klian Intan, and Sungai Lembing were analysed to identify distinct mineralization patterns. The results revealed that the integration of Autoencoders with K-Means yielded the highest clustering quality, with a Silhouette Score above 0.4 and a Calinski-Harabasz Index of 90 at four clusters, outperforming the other methods. The classification effectively distinguished between Pegmatite, Hydrothermal Veins, Polymetallic, and Disseminated deposits, aligning with the geological characteristics of the regions. These findings enhance the understanding of tin deposit distribution, offering significant potential for optimizing exploration strategies and mining operations, thereby contributing to the sustainable economic development of Malaysia's tin mining industry.
Keywords
tin deposits; clustering techniques; geochemical analysis; Malaysia; mineral exploration
Hrčak ID:
333281
URI
Publication date:
3.7.2025.
Visits: 648 *