A Simulation-Based Investigation of Unbiased Detrended Fluctuation Analysis

Poster presenting a large simulation study comparing Unbiased Detrended Fluctuation Analysis (UDFA) with Evenly Spaced DFA (EDFA) and Overlapping-window DFA (ODFA) across a broad range of scaling exponents and time-series lengths representative of biomechanical and physiological data. Contrary to the method’s introducing study, the results indicate that ODFA and EDFA generally outperform UDFA except under strong negative autocorrelation, and that all three methods exhibit systematic bias for short series. Methodological companion to the short-series DFA comparison applied to psychological and gait-synchronization time series presented at SCiP 2019 (Montreal).

Sep 4, 2020

Fifth Annual Conference in Human Movement Variability & First Annual Great Plains Biomechanics Conference