Time series analysis, or the quest for quantitative measures of time dependent behavior

Alfonso M. Albano1*, Peter D. Brodfuehrer1, Christopher J. Cellucci2, Xenia T. Tigno3, and Paul E. Rapp2
1Bryn Mawr College, Bryn Mawr, Pennsylvania, USA
2Uniformed Services University, Bethesda, Maryland, USA
3University of South Florida, Tampa, Florida, USA




We review some of the tools used to analyze time series—sequences of measurements of one or more variables pertaining to the same system. The tools include time-resolved statistics and spectra; autocorrelations, mutual information and transfer entropy; complexity, predictability and hypothesis testing. We illustrate the use of these tools by analyzing extracellular recordings from the leech, vasomotion data from monkeys, and human electroencephalograms.

*To whom the correspondence should be addressed.
Received October 21, 2008; Accepted November 1, 2008.
Editor-in-charge: Amador C. Muriel