Geochemical Anomaly Detection in the Irankuh District Using Hybrid Machine Learning Technique and Fractal Modeling

Prediction of elemental concentrations is essential in mineral exploration as it plays a vital role in detailed exploration. New machine learning (ML) methods, such as hybrid models, are robust approaches infrequently used concerning other methods in this field; therefore, they have not been examined properly. In this study, a hybrid machine learning (HML) method was proposed based on combining K-Nearest Neighbor Regression (KNNR) and Random Forest Regression (RFR) to predict Pb and Zn grades in the Irankuh district, Sanandaj-Sirjan Zone.. The aim of the proposed study is to employ the hybrid model as a new method for grade distribution. The KNNR-RFR hybrid model results have been applied for the Pb and Zn anomalies classification. The hybrid (KNNR-RFR) method has shown more accurate prediction outputs based on the correlation coefficients than the single regression models with 0.66 and 0.54 correlation coefficients for Pb and Zn, respectively. The KNN-RF results were used to classify Pb and Zn anomalies in the study area. The concentration-area fractal model separated the main anomalous areas for these elements. The Pb and Zn main anomalies were correlated with mining activities and core drilling data. The current study demonstrates that the hybrid model has a substantial potential for the ore elemental distribution prediction. The presented model expresses a promising result and can predict ore grades in similar investigations.