ABSTRACT
Optimal mining selection considered in terms of the expected profitability/loss involves post-processing conditional probability distributions of grade values using utility (economic objective) functions. This technique relies implicitly on the conditional probability distributions reflecting adequately the associated spatial uncertainty in grade values. This paper firstly explores three characteristics of loss functions that may be potential problems in their practical implementation for mining category selection. To explicitly account for poor results when using loss functions a maximum profitability with minimum risk framework for ore/waste classification is proposed. This new approach to profitability-based ore/waste classification is based on economic objective functions with global constraints and is demonstrated with an artificial dataset.
