Guided Interactive Visualization for Detecting Cyclical Patterns in Time Series

de Luz Palomino Valdivia F.; Baca H.A.H.; Solis I.S.; Cruz M.A.; Valdivia A.M.C. · Lecture Notes in Networks and Systems (2023) · DOI: 10.1007/978-3-031-28076-4_29

Resumen

Different analytical methods can be applied to find cyclical patterns in time series, so choosing the correct method is not always an easy task; this is where we consider the application of Visual Analytics (VA) that facilitates reasoning, interaction, and search for patterns; for this reason, we have built an interactive visual tool that allows us to perform a univariate and multivariate visual analysis of cyclical patterns in climate time series in a guided manner. A spiral visualization has been used for the univariate data analysis, and star-type glyphs have been used for the multivariate data analysis. The type of guidance applied is orientation based on visual signals to make it easier for the user to enter the number of segments, the main parameter that facilitates the detection of cyclical patterns; for this, the Fourier Transform was used. Three questions were raised to validate the tool according to the objectives outlined.

Instituciones: Universidad Nacional de San Agustín de Arequipa; Universidad Nacional Micaela Bastidas de Apurímac; Universidad Nacional José María Arguedas