<p>The identification and interpretation of geochemical anomalies are fundamental tasks in geochemical exploration. While unsupervised machine learning algorithms are extensively employed for detecting geochemical anomalies, the results often present significant interpretative challenges, complicating subsequent geochemical verification. To address this issue, a sequential machine learning framework designed to both identify and interpret integrated geochemical anomalies is introduced in this study. Initially, the isolation forest (IF) algorithm was applied to 17 geochemical elements to detect integrated anomalies. The extreme gradient boosting algorithm was subsequently employed to perform a regression analysis and model the relationships between the 17 geochemical&#xa0;elements and the IF-derived anomaly scores. Then, the Shapley additive explanations (SHAP) method was used to interpret these anomalies at both global and local scales and to delineate target areas for potential mineral deposits. The results indicate that the IF algorithm can be effectively implemented to identify geochemical anomalies associated with mineralization. Global SHAP value analysis suggests that the predominant mineralization type in the area is likely medium- to high-temperature magmatic–hydrothermal mineralization, with indications of deep-seated material contributions to the metallogenic process. The primary ore-forming elements identified include W, Cu, Pb, and Zn. By integrating the geochemical anomalies with SHAP-derived insights, nine prospecting targets were delineated. On the basis of field verification and prior exploration data, these areas are inferred to hold significant potential for Zhuxi-type W–Cu polymetallic deposits. This research presents a robust machine learning framework for the identification and interpretation of geochemical anomalies, offering valuable guidance for mineral exploration in the region.</p>