Abstarct: The high dependence of knowledge-driven mineral potential modeling methods on expert knowledge and experience makes the results often erroneous and not unique. Conventional knowledge-driven methods usually use expert knowledge and experience in different stages of modeling, including the preparation of conceptual model, determination of important variables, weighting to various classes of evidence maps or information laxyers, etc. This problem, along with other factors such as the need to determine the type of mineralization sought before modeling, makes it difficult to use these methods in mineral exploration. In this study, principal component analysis (PCA) and independent component analysis (ICA) are introduced as two knowledge-driven methods with the least reliance on expert knowledge and experience to model mineral potential. These methods due to the multivariate structure, low volume of calculations, and lack of need for basic information are primarily consistent with the nature of mineral resource exploration studies and the data used in them. Also, due to the lack of need to determine the type of mineralization sought before modeling, obtaining various results from the implementation of a single modeling process and reducing the impact of expert judgment in the implementation of modeling, have shown positive efficiency in mineral potential modeling. In the present study, in order to develop these methods, after making the necessary preprocessing, the data were entered directly into PCA and ICA process. Then, according to the load values of the variables in the table of PCA eigenvectors and ICA unmixing matrix and examining the positive and negative correlation of variables in these tables, each of the output components has been interpreted. In order to implement this process, an area of 4800 square kilometers in the south of Neishabour, northeast of Iran has been used. High-quality data such as the results of geochemical stream sediment, geological information, structural fracture patterns, and remote sensing data are available in this area. After modeling, two different types of mineralization including podiform chromite deposits and epithermal gold-antimony deposits were identified in this area. Then, the accuracy of the results obtained from the two methods was evaluated using the position of the known mineral prospects in the region and the method of receiver operating characteristic curve. The obtained results were also compared with the results of univariate and multivariate geochemical studies and the results of mineral potential modeling by the fuzzy logic method. In addition, the effect of outlier values correction on ICA and PCA modeling results was also studied. The results of this study showed that PCA and ICA modeling were relatively successful and identical in identifying promising areas of podiform chromite deposits and epithermal gold-antimony deposits in the study area. The ICA method has yielded more accurate results due to more rigorous theoretical principles and the use of higher-order statistics. However, the area of anomalies introduced by it was smaller and more incomplete than the PCA method. In general, it can be said that ICA is more accurate in identifying promising areas for mineralization, and in contrast, PCA has introduced more comprehensive anomalies, despite the small error in identifying promising areas for mineralization. Also, the introduced methods have been successful in using less expert judgment and have used expert knowledge and experience only in the stage of interpreting the results. In addition, the two methods used without prior information from the study area, have succeeded in identifying both types of mineralization in the area. These two cases can be effective steps towards easier and more effective use of knowledge-driven methods of mineral potential modeling in exploratory studies.