Time-Series Descriptor

Processing steps: a time series curve, its piecewise aggregate approximation, a string of letters and a color matrix

Time series provide massive data of multiple input sources. The implemented descriptor helps to reduce the amount of data and allows for better comparison and visualization. It uses the Piecewise Aggregate Approximation (PAA) to reduce the amount of data and the Symbolic Aggregate approXimation (SAX) to transform the data into a symbolic representation. Finally, the project displays the representation of the series as a color-matrix to the user.

Screenshot of the tool comparing two time series as color matrices side by side

Screenshot of the tool showing a large color-coded matrix of the full time series data

The time-series descriptor was developed together with Jan Riemann and Michael Stoica. It was part of the seminar Visual Analytics at Fraunhofer IGD and TU Darmstadt during the summer term 2010. The project was supervised by Dr. Tobias Schreck and Jürgen Bernard.