ANÁLISE DE AGRUPAMENTO APLICADA AO RECONHECIMENTO DE DIFERENTES TIPOS DE ITABIRITOS SILICOSOS FRIÁVEIS DA MINA DE BRUCUTU

Cluster analysis is a multivariate statistical technique that allows to separate the sample units into internally homogeneous and externally heterogeneous groups. In this study the cluster analysis was applied through R software, with the aim of stratifying into subgroups the ores classified as friable siliceous itabirite from the Brucutu Mine, Quadrilatero Ferrifero, state of Minas Gerais, Brazil. The data matrix consists of 3519 samples and 10 variables related to granulometric and chemical data. The hierarchical clustering analysis showed that the samples could be subdivided into four groups, and then they were grouped by the non-hierarchical K-means method. Through the analysis of the features of each group formed they were classified and related to the geology of the deposit. Typology 1 (friable, Fe-rich and fine siliceous itabirite) is located at the top of the deposit and stratigraphically above the intrusive mafic rocks. Typology 2 (– friable, low grade and coarse siliceous itabirite) is marked by NW-SE direction, Brasiliano event. Typology 3 (friable, low grade and fine siliceous itabirite) is dispersed throughout the deposit, making it difficult to correlate with some geological phase. Typology 4 (low grade semi-compact itabirite) is demarcated by concentration in the SW region of the reservoir.