A comparative analysis between the Geostatistics and Machine Learning methods for mineral resource estimation

Abstract

It is presented in this article two methods for estimating minerals resources. The first one is well known in the literature of mining engineering, it is called Ordinary Kriging, which is one of most used algorithms in Geostatistics to deal with regionalized variables. The second one is not yet so popular to be used for estimating minerals resources, although its applicability to be very effective for this task. This method is called Random Forests, which is an algorithm of Machine Learning. Thus, its analysis will focus on the construction procedure of block models of a copper mineral deposit using both methodologies.