In this study, the relationships between sensor detected features and off-line analysed metal yields during hydrometallurgical processes are described using data-driven models. The captured on-line features are used as indirect indicators to predict the metal yields during the experiments. The predictions exhibit a commendable degree of consistency with the calculations from offline analysis. The proposed method proved the feasibility of employing data-driven techniques in hydrometallurgical recycling processes. Furthermore, the method can be used as a surrogate model for offline chemical analysis and provide timely and informative analytic insights.