Klasterisasi Kadar Endapan Nikel Laterit Berdasarkan Data Lubang Bor Menggunakan Metode K-Means Clustering

Abstract

Laterite nickel deposits are one of the mineral commodities that play an important role in the mining industry. Exploration drilling is the initial activity in nickel laterite mining and produces a distribution of drill holes with high Ni and Fe grades and complex spatial patterns. This distribution can be identified using machine learning, particularly clustering, to classify drill holes based on Ni and Fe content. In this study, unsupervised learning using the k-means clustering method was applied to identify drill hole distribution patterns. In addition, this study also analysed differences in characteristics between laterite zones. The data used in the study consisted of 291 drill holes, containing Ni and Fe grades as well as limonite and saprolite zone lithology from exploration drilling. The results indicated that clustering based on Ni and Fe grades produced two clusters in the limonite zone and three clusters in the saprolite zone. The linear relationship between Ni and Fe in the limonite zone showed a significant negative correlation (r = –0.32), whereas in the saprolite zone it showed a very weak correlation (r = 0.11). The difference in Ni content between the limonite and saprolite zones was significant (p-value < 0.05), while the difference in Fe content was not significant (p-value > 0.05). Spatial analysis showed that clusters with higher Ni grades were dominant in the saprolite zone, whereas Fe grades were dominant in the limonite zone. Based on these results, the k-means clustering method was able to effectively group drill hole data based on Ni and Fe content and highlight differences in Ni and Fe characteristics between the limonite and saprolite zones