Generalized numerical lattices for time series representation in complex data systems
Thalita Veronese · Journal of Computational Interdisciplinary Sciences · 2009
Analysis of information from multiple data sources obtained through high resolution instrumental measurements has become a fundamental task in all scientific areas.The development of expert methods able to treat such multi-source data systems, with both large variability and measurement extension, is a key for studying complex scientific phenomena, especially those related to systemic analysis in space and environmental sciences.In this paper, we propose a time series generalization introducing the concept of generalized numerical lattice, which represents a discrete sequence of temporal measures for a given variable.In this novel representation approach each generalized numerical lattice brings post-analytical data information.We define a generalized numerical lattice £ as a set of three coefficients (κ, λ ℓ , μ p ), representing the following data properties: dimensionality, size and post-analytical parameters, respectively.From this generalization, any multi-source database can be reduced to a closed set of classified time series in spatio-temporal generalized dimensions.As a case study, we show a preliminary application in space science data, highlighting the possibility of a real time analysis expert system to be developed in a future work.