Scholarly record

FUZZY CLUSTER ANALYSIS TO IDENTIFY ORE TYPES IN KHOUNI AREA, IRAN

A. Pourjabbar, A. Hezarkhani, S. A. Meshkani

Abstract

Multivariate statistics analysis is one of the base stages in geochemical data processing. In this branch of mathematics, “Hierarchical Clustering Analysis” applied to identify the relations between the variables such as elements and samples. Hence, different types of mineralization would define. However, any missing or errors in entrance the data make a wrong or not clear result. Sometimes misinformation or data errors appear because of analytical errors, missing some information, or problems in sam pling or sample preparing. These errors have bad effects on the statistical studies. In this situation, interpreting the results is hard or wrong decision would make. Former studies showed that “Fuzzy Cluster Analysis” could apply in these situations successfully. In this paper, the “Fuzzy Cluster Analysis”, applied to classify the anomalous samples of Khouni Mountain, Central. Iran. 254 stream sediment samples applied for this purpose. They have analyzed for 17 elements by ICP method; and 90 anomalous samples identified by “×− + ns” method, a classic statistical technique. These samples classified into groups based on the “Fuzzy” and “Classic” classification. For each method, a dendrogram plotted to show the result more obviously. Comparing these two methods, show that both of them have the same result. The two diagrams are like each other at the first glance; but the “Fuzzy dendrogram” has a higher significant level. However, there are some misclassifications in “Classic Diagram”. In this dendrogram, some samples don’t have meaningful relationship with other ones in the same group. These wrong sites make it difficult to interpret the result. To avoid such problems, fuzzy method is suggested.

Publication details

Back to publication list