<p>Anomalies reflected in 1691 soil samples from a copper (Cu) porphyry deposit in British Columbia province, Canada, have been investigated and the primary promising areas of the soil samples identified. Also, the data of 83 rock samples collected from the previously identified areas have been analyzed in detail. To analyze the rock data, the <i>F</i> test feature selection method was used to realize and rank the important features so that the five elements of Au (gold), Mo (molybdenum), Ag (silver), Fe (iron), and Pb (lead) were selected as the important features to be applied for running the regression models. Furthermore, 70% of the data were chosen for training and the rest for testing the model. The methods of the linear regression model, regression tree model, ensemble bagged trees model, ensemble boosted trees model, and the three-layered neural network model were used first, and then the data were analyzed using the new proposed support vector regression (SVR) model. The <i>R</i>² value of the proposed SVR regression model with five selected features was the best for the validation and test data and equal to 66% and 77%, respectively, among all the regression models. In addition, the RMSEs for this model for the validation and test data were 0.0879 and 0.1045, respectively, which are the least among all the models, while the training time for this model was 1.34 which was less than for the other models. Finally, the maps for the predicted values of Cu were drawn and the results of all methods were compared, indicating the efficiency and superiority of the proposed method.</p>