Rock hardness and copper recovery prediction using textural clustering

•This paper provides a methodology for the combination of different sources of data for geometallurgical modelling.•Qualitative and quantitative datasets with different spatial resolution were successfully combined.•The dataset was used to predict copper recovery and rock hardness.•The accuracy increase by the incorporation of textural information from core hyperspectral imaging was tested.•Results suggest that in this dataset, textures are relevant to recovery, but not to hardness.•The paper highlights the challenges in data quality and its handling for the development of predictive models.