Application of K-Means algorithm to classify geochemical populations in Vein-type copper deposit in ۱:۱۰۰,۰۰۰ sheet of Kariz now, Khorasan, Iran

TO address the challenge of limited access to mineral resources, this research utilizes a combination of dimensionality reduction techniques and sophisticated machine learning methods to significantly improve the accuracy and speed of detecting geochemical anomalies within stream sediments. The focus is particularly on the Razavi Khorasan province in northeastern Iran. The approach begins with Principal Component Analysis (PCA), applied after initial data preprocessing and the creation of a catchment basin incorporating associated values. Following PCA, the study employs K-Means, an unsupervised clustering algorithm, to categorize geochemical values into meaningful groups. This methodology uses preliminary data processing via correlation matrices and dimensionality reduction, followed by application of K-Means clustering on centered log-ratio transformed values of chosen elements. Through this computational model, the research effectively classifies stream sediment samples into three distinct clusters, offering a deeper insight into the areas geological features and the identification of possible mineral reserves responsing te the need to understand geochemical halos in a more accurate way.

کلیدواژه ها:

K-Means clustering algorithm ، Unsupervised machine learning ، silhouette score ، elbow method ، PCA