The Prediction Of Mineral Resources Based On Data Mining Techniques

In recent years, Data Mining techniques can obtain useful information from large, high-dimensional, fuzzy and noisy data, which provides a theoretical basis for efficient, rapid and scientific process of geochemical data.In this paper, Data Mining techniques are used in prediction of regional mineral resources. Firstly the data which is geochemical exploration data of39kinds of elements of1:200000international division K-49-(20) is sorted into Excel file and TXT file. The arranged data is made to K-means Cluster Analysis by PolyAnalyst which is a data mining software. Then six areas are found, which are relatively stable and have same variation characteristics of geochemical elements. In order to test the effect of cluster analysis, the line charts of39sorts of elements’ content of different sampling points in each region are mapped, it is found that geochemical elements’ content of different sample point have almost the same characteristics in each one of selected six regions and geochemical elements’ content of different regions is obviously different. It shows that the result of clustering is effective. By mapping the line charts of target elements’ content of common mineral resources in the six regions, therefore the abnormal elements in every area are determined. Then it is found the four of six regions selected exist possibly mineral resources and forecast possible types of mineral resources preliminarily. Secondly using the Correlation Analysis technique analyze the geochemical elements of four regions, and find out the correlations between elements. Then predict existed possibly mineral resources types in different regions on the basis of geochemistry theory. Getting intersection elements between the predicted results of Correlation Analysis in four regions about mineral resources and the result of preliminary prediction, I inferred mineral resources exist possibly of four regions. For example, sulfide deposit, precious metal vein deposit, skarn deposit, ultramafic rock deposit, porphyry copper deposit or vein type uranium deposit. Finally, these four regions are painted on the geological map as target regions by MapGIS K9software.Comparing the predicted results and actual conditions, two regions of the four predicted ones have been found to exist mineral resources, and in some part of the others also have been found in the favorable metallogenic zone. The research shows that the data mining technology can be used for the prediction of mineral resources effectively.