<p>Identification of mineralization-related geochemical anomalies is a key step for geochemical exploration in an area with a complex geological background. In this study, a combined model integrating robust factor analysis and robust regression analysis was developed for distinguishing the stream sediment geochemical anomalies. The concentrations of major rock-forming oxides were applied to a robust factor analysis for identifying element associates and principal factors representing different lithological types. Subsequently, these factors were used as independent variables in a robust regression analysis to model the background variations of trace metals. The resulting geochemical residuals, representing the discrepancy between measured and predicted values, were then defined as new geochemical exploration indicators. The 1:200,000 stream sediment geochemical data in the Yinkeng Orefield were analyzed, and the results revealed that the lithologic background is characterized by the first three principal factors, including F₁ (Al₂O₃-Na₂O-K₂O), F₂ (CaO-MgO) and F₃ (MgO-Fe₂O₃), which represent intermediate–acid magmatic plutons, carbonate and other calcium-magnesium sedimentary rocks, and basic–ultrabasic intrusive rocks, respectively. Among the six trace elements (Au, Ag, Pb, Zn, W, Sn), the concentrations of Pb and Sn show strong positive correlation with F₁, whereas those of Zn are influenced by both F₁ and F₃. Compared to the measured values, the residuals more effectively identify mineralization-related anomalies, as evidenced by a stronger spatial correlation between high residual areas and the known ore deposits, indicating that the proposed model is a viable technique for geochemical exploration in geologically complex areas.</p>