Statistics > Methodology
[Submitted on 5 May 2020 (this version), latest version 26 Jul 2021 (v3)]
Title:Finite Sample Smeariness of Fréchet Means and Application to Climate
View PDFAbstract:Fréchet means on manifolds are minimizers of expected squared distance, just as their Euclidean kin. On manifolds, Fréchet means for most distributions feature a classical asymptotic Gaussian central limit theorem with $\sqrt{n}$-rate, except for some distributions, some of which are known on spheres. They feature slower and non-Gaussian rates, called \emph{smeary} rates. Here we first describe exhaustively all smeary distributions on circles, comprising not only \emph{power} smeariness, but also \emph{logarithmic} smeariness. Then we make the concept of \emph{finite sample smeariness} (FSS) precise and show for the circle that FSS affects all distributions that are spread on both sides beyond an open half circle. An analog results holds for tori. We statistically quantify the \emph{scale of FSS} and find two different qualities: If asymptotically, the sample Fréchet variance remains strictly above the population Fréchet variance, we speak of Type I FSS, this is the case, e.g. for all (!) von Mises distributions. Else, we have Type II FSS, e.g. for distributions with support excluding the antipode of its mean. For both types of FSS it turns out that the nominal level of nonparametric asymptotic quantile based tests for the circular mean is higher than the true level. Simulations indicate, however, that suitably designed bootstrap tests preserve the level. For illustration of the relevance of FSS in real data, we apply our method to directional wind data from two European cities. It turns out that quantile based tests, not correcting for FSS, find a multitude of significant wind changes. This multitude condenses to a few years featuring significant wind changes, when our bootstrap tests are applied, correcting for FSS.
Submission history
From: Stephan Huckemann [view email][v1] Tue, 5 May 2020 16:28:35 UTC (322 KB)
[v2] Fri, 29 May 2020 07:04:16 UTC (704 KB)
[v3] Mon, 26 Jul 2021 20:51:35 UTC (1,808 KB)
References & Citations
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.