engleski

Classification of Hydrochemical Data in Reduced Dimensional Space

The main objective of this paper is to systematically analyze the performance of water sample classifications for different data representations. We compare the classification of full data representation to the classification of data items in a lower dimensional space obtained by the projection of the original data on the space formed by first principal components, and further, on the space of centroids of classes. We use linear support vector machines for classification of ground water samples collected from five different localities of the Odra River basin. The obtained results are evaluated by standard measures including recall, precision and F1measure.

concept decomposition; dimensionality reduction; principal components analysis; support vector machines

nije evidentirano