Abstract: Metallogenic prediction,a core component of mineral resource exploration and evaluation,is gradually evolving toward quantification and intelligence,driven by the rapid development of big data and machine learning technologies.To address the difficulty that a single metallogenic prediction model strugglest to balance accuracy and interpretability,this paper systematically reviews the basic principles,modeling workflows and applicability of two traditional statistical methods,the weight of evidence(WoE)and the fuzzy weight of evidence(fuzzy WoE)) as well as two machine learning algorithms,random forest(RF) and support vector machine(SVM).By introducing accuracy evaluation metrics such as ROC curve,AUC value and Kappa coefficient,a multi-source information fusion metallogenic prediction model coupled with traditional methods-machine learning is constructed,and an empirical verification is carried out by taking a coal mining area in Shanxi Province as the study area.The results indicate that the traditional WoE method offers strong interpretability but has limited nonlinear fitting ability;the fuzzy WoE method improves the adaptability to uncertain information through fuzzification processing,and its prediction accuracy is 5%～10% higher than that of the traditional weight of evidence method;RF and SVM achieve higher accuracy in processing high-dimensional nonlinear geological data but exhibit a "black box" characteristic.In contrast,the coupled model,constructed on the basis of traditional methods and machine learning realizes the organic integration of the advantages of various methods.In this paper,a 1:1 balanced set of positive and negative samples was established,and a spatial buffer avoidance strategy was adopted to select non-mineral samples.Model parameters were optimized by means of grid search,and verification was ultimately conducted on 962 grid units in the study area,including 48 known mineral sites and 48 inferred non-mineral sites.The results show that the coupled model reaches a prediction accuracy of 90.1%,an AUC value of 0.91 and a Kappa coefficient of 0.75,and its prediction performance is significantly superior to that of single models.The construction of a collaborative prediction system through multi-method fusion can effectively improves the reliability of mineralization prediction,provides technical support for mineral exploration under complex geological conditions,and lays a theoretical foundation for establishing a multi-method integrated mathematical geological prediction framework.