What is it about?
It is shown how ordinal patterns can be used for distribution-free hypothesis tests, where the null hypothesis of serial independence in real-valued time series is tested against (possibly non-linear) serial dependence. We provide simple closed-form expressions relying on asymptotic derivations for implementing these tests. The performance of these tests is investigated with simulations, and their usefulness is illustrated by an environmental data example.
Featured Image
Photo by Jake Hills on Unsplash
Why is it important?
The use of ordinal patterns leads to a distribution-free method, which is attractive for applications in practice. As simple closed-form asymptotics are derived for the considered statistics, a computationally efficient implementation of the resulting hypothesis tests is possible.
Perspectives
To me, this article is the starting point for several follow-up research topics, such as adaptions to statistical process monitoring (sequential testing via control charts) or adaptions to discrete-valued time series (modified ordinal patterns).
Christian Weiß
Helmut-Schmidt-Universitat Universitat der Bundeswehr Hamburg
Read the Original
This page is a summary of: Non-parametric tests for serial dependence in time series based on asymptotic implementations of ordinal-pattern statistics, Chaos An Interdisciplinary Journal of Nonlinear Science, September 2022, American Institute of Physics,
DOI: 10.1063/5.0094943.
You can read the full text:
Contributors
The following have contributed to this page
