Machine learning model to assess gold mineralization potential mapping for northwest Thanh Hoa province

Abstract

In Vietnam, gold is one of the key minerals for mining. While gold mining in Thanh Hoa province holds economic value, evaluating and predicting its spatial distribution remains challenging. Machine learning is becoming a powerful tool in the field of mineral research and extraction, including gold. The strength of Machine Learning lies in its ability to process and analyze large amounts of data to make more accurate predictions, optimize exploration and extraction processes, and minimize risks. This paper presents a set of machine learning models to identify the best model for generating a gold deposit potential map. The study utilized seven thematic maps, including lithology, magnetic data, gravity data, geological age, faults, lineaments, ore point density, magma distribution, gold mineral potential from remote sensing imagery, and placer gold distribution, as input data for the models. Additionally, 706 points (353 sampling sites with gold placer and 353 sites without gold) were used to generate the training and testing dataset. The study area is situated in the northwestern part of Thanh Hoa province, known for its high potential for gold deposits. Machine learning models such as Random Forest, Logistic Regression, SVM, and Gradient Boosting were implemented to identify the best-fit model for the study area. After comparing the models, the initial results showed that the Random Forest model achieved the highest accuracy with an AUC of 0.82, identifying 4% of the area as having very high potential. The final result, a gold mineral potential map, was compared with field data, and it was found that all points containing gold in the field were located in the areas (very high potential) in the prediction map.