Due to the discovery of shallow ore deposits, the exploration of the subsurface and deep-ore reserves has flourished. Gold is a precious rare element which has been heavily taken into account in recent years due to the supply and demand market. One of the fundamental issues in gold exploration is the consideration of temporal and financial optimization. Analysis of geochemical data has become an important task due to a systematic sampling of soil and rock, the study of the relationship between elements, and the identification of high mobility elements. In order to discover knowledge from data, an efficient tool titled Data Mining has been used. Therefore, instead of designing a drilling network based on the surface data, data processing techniques are used to specify the optimal drilling points. This process reduces the risk as well as the financial and temporal costs. In this thesis, basic and advanced statistical techniques have been used to model the Chah-e-Zard (low to intermediate sulfidation) epithermal Au-Ag deposit. The mineralization type is associated with Miocene volcanic intrusive bodies along with calc-alkaline magma, andesite, trachy-andesite, and dacite. The modeling process is presented as Absolute Error of 5%). In the prediction models, in order to estimate the each of the defined parameters, multiple linear regression, Multilayer Perceptro(MLP) and Radialasis Function (RBF)eural network, Support Vector Regression (SVR) and model tree technique are used. Accordingly, the model produced by the M5P model tree has a better performance compared to other methods (with an adjusted Coefficient of determination of 88% and Mean Absolute Error of 11%). The results of classification and prediction models indicate the ability of decision tree techniques to provide optimal and low-risk solutions. Based on the generalizability of these models, additional drilling points were proposed in the undrilled areas in the western part of the region.
