Estimating 1/fα scaling exponents from short time-series

In recent years, there has been a concerted effort to develop methods for estimating the scaling exponents of time-series data, thus permitting a characterisation of their underlying dynamical behaviour. This task becomes rather inaccurate with data of limited length (less than 100 points), as is the case in many real studies where observation time is constrained. In this paper, we explore a novel method for accurately calculating the scaling exponents of short-term data, using what we term the multiple segmenting method (MSM). This approach relies on maximising the available information within a time-series by generating pseudo-replicates. We believe this method is potentially useful, especially when applied to biological data.