http://www.eckner.com/research.html New and updated research papers by Andreas Eckner en-us research.html Today I posted an updated version of my paper that provides a systematic framework for working with unevenly-spaced time series. I finally feel that the paper is ready to be sent out more broadly. Most changes since the March 2012 version are relatively small. The clarity and spelling was significantly improved by hiring Bioscience Writers as a professional proofreader. Sun, 26 Aug, 2012 14:00:00 EST www.eckner.com/4 research.html Today I posted an updated version of my paper on algorithms for unevenly-spaced time series. The changes since the December 2011 version are fairly minor, mainly fixing typos, and bringing explanations and notation in line with my theoretical paper on unevenly-spaced time series ("A Framework for the Analysis of Unevenly-Spaced Time Series Data"). The implementation now illustrates the compilation using the 64 (instead of 32) bit version of GCC under Windows. Thu, 12 Apr, 2012 15:13:00 EST www.eckner.com/1 research.html Today I posted an updated version of my paper on trend and seasonality estimation for unevenly-spaced time series. There are two major changes since the November 2011 version. First, the paper now includes a simulation study of the bias (depending on the degree of irregularity of inter-observation times) in estimating the peak-to-trough size of the seasonal component. Second, I discuss several possible extensions of the paper's algorithm: robustness, time-varying seasonality, and real-time estimation. Tue, 10 Apr, 2012 12:39:00 EST www.eckner.com/2 research.html Today I posted an updated version of my paper that provides a systematic framework for thinking about and working with unevenly-spaced time series. There are many changes since the October 2011 version, especially in the second half of the paper. An entire section is now devoted to linear time series operators and time series sample paths. It shows that - when fairly general conditions are met - there is a one-to-one correspondence between bounded (or equivalently, continuous) linear time series operators and convolution operators. The section on trend and momentum has been shortened and merged into the section on moving averages, since I deemed the results less important and in order to shorten the paper. Finally, the section on scale and volatility has been completely revised and now provides rigorous proofs for the consistency of various volatility estimators. Wed, 28 Mar, 2012 20:13:00 EST www.eckner.com/3