Abstract This study applies a D-vine copula modeling approach to predict metallurgical recovery (Rec) from three key geometallurgical variables: semi-autogenous grinding (SAG) power index (SPI), resistance to abrasion and breakage index (A*b), and Bond work index (BWi). A dataset comprising 775 diamond drill core samples from Tarkwa Mine, Ghana, was used to develop the model. The marginal distributions of the variables were combined through copula-based quantile functions to construct a predictive framework, which can be integrated into mine planning and scheduling. The model achieved an average predicted recovery of 89.26%, closely matching the observed average recovery of 88.64%, with a mean absolute error (MAE) of 4.41, indicating satisfactory performance. These results demonstrate that the proposed copula-based approach offers a robust alternative for modelling geometallurgical variables and supports data-driven decision-making in mining operations.
Adusei et al. (Sun,) studied this question.