A data-driven system for mineral exploration decision-making using value of information and Modern Portfolio Theory

Abstract:

A problem in mineral prospectivity mapping is where to collect additional data once an area has been targeted for exploration. “Boots on the ground” exploration involving a drilling campaign or geochemical sampling represents a major commitment to an exploration prospect compared to methods involving geophysical surveys or “big data”. It is in a mineral exploration company’s best interest to commit to projects that have the highest probability of success, or else waste considerable time, effort, and exposure (safety, environmental, legal) to inappropriate levels of risk. We present a method based on Modern Portfolio Theory that informs mineral explorers where to collect additional data that maximizes the total value of the exploration project. The mineral exploration decision-making workflow is quantified using value of information, which reduces bias introduced during subjective appraisals of exploration success. Quantification allows for the mineral exploration process to proceed in a data-driven manner, where many hypotheses can be rapidly tested and assess uncertainty. We incorporate the concepts of Modern Portfolio Theory, gambler’s ruin, and multi-armed bandits in this novel approach, and demonstrate results on a real dataset located in the Gawler Craton of South Australia.