Exploring chaotic time series and phase spaces:from dynamical systems to visual analytics

The analyses of time series can be enhanced by modeling it in the phase space, where its dynamics are described in a more intuitive manner. In this context, time-series observations are organized in the form of states, which are (hopefully) bounded by a well-defined structure where better insights can be inferred. However, two parameters are needed to perform such a transformation. Based on the limitations of current methods to estimate those parameters, this Ph.D. thesis investigates more robust alternatives to reconstruct phase spaces from time series