TEStool: A Visual Interactive Environment for Modeling Autocorrelated Time Series

TEStool is a visual interactive software environment for modeling autocorrelated time series, using a versatile class of stochastic processes called TES (Transform-Expand-Sample). The novel feature of the TES modeling approach is that it strives to t a model to empirical records by simultaneously capturing both the empirical distribution and the leading empirical autocorrelations. Thus, TES models can have a high degree of delity, since both rst-order and second-order aspects of the statistical signature of empirical time series are targeted for capture. TEStool has been used extensively to model empirical data from a variety of application domains, including compressed video traac in high-speed communications networks, nancial time series and machine reliability. This paper explains the TES approach to modeling empirical autocorrelated time series adopted in TEStool, as well as TEStool's graphical user interface (GUI). Special emphasis is placed on features that increase the eeciency and quality of modeling based on visualization and real-time interaction. In particular, TEStool casts a heuristic search activity into an intuitive process which is guided by visual feedback. Consequently, the software makes TES modeling accessible to modelers who have a basic understanding of time series, but lack expert knowledge of TES theory.