Estimation of information contained in soil attributes

Abstract

In classification tasks the information contained in particular soil attributes denotes the relative contribution of each of the attributes into soils, horizons and samples discrimination. The bigger such contribution of a given attribute is, the more information the attribute contains. High variation of its values can mask the heterogeneity of the data set, which may potentially consist of several groups of objects. At the same time there are more complex multivariate estimates with  the properties required for criteria, which use not only attribute variation, but also their interrelationship. These methods are principle components analysis, multivariate analysis of variance and cluster analysis. Formalization of concepts, definitions and classification techniques is the most pressing issuein soil science, as the mere listing of soils should be substituted by the approach based on informatics technologies and mathematics. Quantitative estimation of the information contained in soil attributes is the primary task in this direction.