Spatial predictive mapping of lithium and rare earth element pegmatites using the non-structural fuzzy decision support system: Example from the Orange River Pegmatite Belt, South Africa

• Spatial predictive maps generated on the non-structural fuzzy decision support system (NSFDSS) model results in accuracies from 81.7% to 82%. • Variance inflation factor and Pearson correlation coefficient found low correlation between predictive variables. • Structural and gravimetric lineaments are associated with zones of the highest prospective areas for lithium and Rare Earth Element bearing pegmatites. Spatial predictive mapping is applied to lithium (Li) and rare earth element (REE) bearing pegmatites in the Orange River Pegmatite Belt of South Africa, through a multi-criteria predictive model utilizing five exploratory datasets for mineralized pegmatite occurrence. To assess data redundancy, the variance inflation factor (VIF) and Pearson correlation coefficient (PCC) were applied to assess statistical relationships among exploratory input data. To examine the relationship between the exploratory data and known mineral occurrences, a prediction area (P-A) plot analysis was conducted. To generate mineral predictive maps for mineralized pegmatites, the non-structural fuzzy decision support system (NSFDSS) model was employed, where receiver operating characteristic (ROC) and area under the curve (AUC) analyses were used to assess their accuracy. VIF and PCC indicated low correlation among predictive variables, while P-A plot analysis showed moderate to strong spatial associations between target and exploratory data, ranging from 0.52 for alteration data to 0.74 for distance to host rock. The NSFDSS model identified high-potential zones for Li and REE bearing pegmatites in the southeastern and central parts of the study area. Validation using ROC/AUC analysis demonstrated the model’s accuracy, with the various model types; semantic model, priority model, normalized model, and weighted normalized model achieving 81.7%, 82.0%, 81.8%, and 82.0%, respectively. The accuracy scores achieved by these models suggests that the NSFDSS approach using these exploratory datasets is effective to support regional scale exploration of Li and REE bearing pegmatites in the Orange River Pegmatite Belt.