Clustering and Principal Component Analysis for the Heavy Metal Contents of Soil in Yunnan Province

Abstract

The clustering analysis method which can extracts the useful as well as potential information and knowledge from the random and quantity data is widely used in many domains as an important task of Data Mining. Single Factor Index, Principal Component Analysis, Hierarchical Clustering Analysis and K-Means Clustering Analysis which were used in the experiment in this paper to reveal the relationships of the 12 species of heavy mental elements including Cr, Ni, Cu, V, Co, Mn, Pb, Zn, As, Se, Hg and Cd in the soil of Yun Nan province. 1781 soil samples of the soil which contain the 12 kinds of heavy mental elements that mentioned above were analyzed in the experiment. These heavy metal elements were separated into four parts by the above clustering analysis methods. The 12 kinds heavy metal elements were divided into 4 groups by the Hierarchical Clustering analysis, one group consisted of r, Co, Cd, Se, Ni, Mn and Hg, one group consisted of Pb and Zn as a cluster, one group consisted of Cu and V, the other was As respectively. As well as 4 parts were divided by the K-means clustering analysis, which is similar with the Hierarchical Clustering analysis, one group consisted of Hg, Cu, Ni, Mn and V, one group consisted of Pb and Zn as a cluster, one group consisted of Cr, Cd and Co, the other was As respectively. The results in this paper present that elements in the same group are strong symbiotic correlation.