<p>In this study, mineral prospectivity mapping (MPM) was conducted to investigate quartz vein-type tungsten mineralization in the Jiaoxi area of the western Lhasa terrane, Xizang, China. Diverse geochemical and geological datasets were compiled and analyzed via the use of advanced machine learning (ML) techniques. Three supervised algorithms, namely, extreme gradient boosting (XGBoost), support vector machine (SVM), and random forest (RF) algorithms, were adopted to integrate 27 evidence layers representing key predictors of tungsten mineralization. As a result, four new tungsten mineralization targets were identified in the Jiaoxi area, which are recommended for drill core testing. The robustness of the generated MPM results was validated based on the accuracy of classification curves, demonstrating the reliability of the models. In addition, unsupervised algorithms, such as principal components analysis, K-means clustering, and t-distributed stochastic neighbor embedding, were applied to investigate linear and nonlinear elemental correlations within the geochemical datasets. The Shapley additive explanations values of the 27 features indicated that the three classification models (RF, SVM, and XGBoost) exhibited high sensitivity to geochemical anomaly clusters (Au, As, Ag, W and F), suggesting that these features were closely related to VEIN-TYPE W mineralization. Moreover, both known and potential tungsten occurrences were effectively classified through the partial least squares–discriminant analysis algorithm, and the second component provided high discriminatory power. These findings highlight the ability of the integration of diverse ML algorithms to provide high-accuracy MPM. This approach offers significant potential to refine exploration strategies and increase the efficiency and sustainability of mineral resource development.</p>