The effect of data standardization in cluster analysis

Abstract

The application of multivariate techniques to experimental results requires a responsibility on behalf of the researcher to understand, evaluate and interpret their results, especially the ones that are more complex. In this work, the impact of three standardization techniques on the formation of clusters by the Kohonen neural network were studied. The techniques studied were logarithm (log10), generalized-log and improved min-max. The studies were performed using two databases consisting of 298 and 146 samples and containing the mass fractions of As, Na, K, La, Yb, Lu, U, Sc, Cr, Fe, Cs, Eu, Tn, Hf and Th, determined by neutron activation analysis. The results were evaluated using validation indices.