Dear R-users and developers,
based on a recent series of of papers [1-6] the package 'multivariance'
(available on CRAN; latest version 2.3.0, 2020-04-23) was developed.
It provides in particular:
+ *fast global tests of independence* for an arbitrary number of
variables of arbitrary dimensions
+ a detection and visualization algorithm for *higher order dependence
structures*
+ estimators for multivariate dependence measures which *characterize
independence*, i.e. the population version is 0 if and only if the
variables are independent (in contrast to the standard correlation 'cor')
As a side remark, some food for thought: Note that in [3] it is referred
to over 350 datasets from more than 150 R-packages, which all feature
some statistical significant higher order dependencies. Some are
probably artefacts, but in any case it is likely that these have been
unnoticed and undiscussed so far. Moreover, since it was purely a brute
force study, this might provide starting points for plenty of research
by the corresponding field specialists.
Comments and questions on 'multivariance' and the underlying theory are
welcome.
Best wishes
Björn Böttcher
References:
[1] B. Böttcher, M. Keller-Ressel, R.L. Schilling, Detecting
independence of random vectors: generalized distance covariance and
Gaussian covariance.
Modern Stochastics: Theory and Applications, Vol. 5, No. 3 (2018) 353-383.
https://www.vmsta.org/journal/VMSTA/article/127/info
[2] B. Böttcher, M. Keller-Ressel, R.L. Schilling, Distance
multivariance: New dependence measures for random vectors.
The Annals of Statistics, Vol. 47, No. 5 (2019) 2757-2789.
https://projecteuclid.org/euclid.aos/1564797863
[3] B. Böttcher, Dependence and Dependence Structures: Estimation and
Visualization using the Unifying Concept of Distance Multivariance.
Open Statistics, Vol. 1, No. 1 (2020) 1-46.
https://doi.org/10.1515/stat-2020-0001
[4] G. Berschneider, B. Böttcher, On complex Gaussian random fields,
Gaussian quadratic forms and sample distance multivariance. Preprint.
https://arxiv.org/abs/1808.07280
[5] B. Böttcher, Copula versions of distance multivariance and dHSIC via
the distributional transform -- a general approach to construct
invariant dependence measures.
Statistics, (2020) 1-18.
https://doi.org/10.1080/02331888.2020.1748029
[6] B. Böttcher, Notes on the interpretation of dependence measures --
Pearson's correlation, distance correlation, distance multicorrelations
and their copula versions. Preprint.
https://arxiv.org/abs/2004.07649
--
Dr. Björn Böttcher
TU Dresden
Institut für Math. Stochastik
D-01062 Dresden, Germany
Phone: +49 (0) 351 463 32423
Fax: +49 (0) 351 463 37251
Web: http://www.math.tu-dresden.de/~boettch/
_______________________________________________
R-packages mailing list
r-packa...@r-project.org
https://stat.ethz.ch/mailman/listinfo/r-packages
______________________________________________
R-help@r-project.org mailing list -- To UNSUBSCRIBE and more, see
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
