An adaptive subspace clustering dimension reduction framework for time series indexing in knime workflows

Abstract

The performance of traditional index structures degrades with increasing dimensionality, concepts were developed to cope with curse of Data Dimensionality [2,5,6] for many domains. Most of the existing concepts exploit global correlations between dimensions to reduce the dimensionality of the feature space. In high dimensional data, however, correlations are often locally constrained to a subset of the data and every object can participate in several of these correlations. Here a novel framework for Knime (Data Analyzing Toolkit)[3] proposed. A subspace clustering is a method which is adopted for the dimension reduction framework. The system produces effective dimension reduction result for the Knime environment and is adaptive for the different data domains. Here each data representations can be plotted as a Tree structure and then can perform subspace extraction which is more helpful for the future pruning activity. The nodes are then transformed to the Knime workflows [3] and can act on different workflow nodes independently. © 2012 Springer-Verlag.