Distributional Data Analysis in Bayes Spaces
Overview
Statistical analysis of data sets whose observations are whole probability distributions, using Bayes space calculus. Distributions (provided as samples, histograms or density functions) are treated as elements of a Hilbert space, so that operations such as mean, variance, distance, principal component analysis and clustering carry over to densities. Reference measures defining the geometry of the space can be chosen and combined. The methodology was first published with peer review by van den Boogaart, Egozcue and Pawlowsky-Glahn (2014) doi:10.1111/anzs.12074.
- Bug fix:
as.BDataDensities()andas.BDataLogDensities()on histogram objects applied the bin weights in reversed order, so that all histogram-derived step densities (and everything computed from them) were mirrored in the domain (.DensityFromHist()/.LogDensityFromHist()). Quantities that are invariant under a coherent relabeling of the domain (variances, distances, depths, explained-variance shares) were only mildly affected through the unequal bin widths, but plotted densities, means, and eigenfunctions appeared reflected. Densities supplied as functions (BDataPDFs,BDataDensities) were not affected. ReferenceMeasureUniform()gains aseedargument: the Monte Carlo sample of the reference measure is drawn under that seed and the global RNG state is restored afterwards (same convention assimulate()). Reproducibility of a reference measure no longer depends on the global seed being set immediately before construction.
Show all changes in 0.1.1Hide the rest of the changes in 0.1.1
plot()onprincomp.BDataobjects withtype = "scores"now acceptspch(recycled across observations), so groups can be distinguished by plotting symbol in addition to color.plot()on multivariate SFPCA objects (type = "eigenfunctions") forwards...to the panel set-up call, so axis labels such asxlabandylabcan be customized.
From NEWS.md in the release
Install
Health
- OK8 OK2026-10-21
Documentation
- Help pages
- 162 (1 internal)includes a package overview page
- Examples that run
- 98%143 run in a plain check · 3 only when donttest examples are enabled (R CMD check --as-cran)
- Exported functions with examples
- 93%
- Documented parameters
- 93%
- Return-value docs
- 98%
- Help pages citing references
- 2%
Downloads
Dependencies
Installing DistributionalDDA adds 44 packages (29 with compiled code, 41 MB of source tarballs). Heaviest: ggplot2 6.0 MB, fds 3.7 MB, compositions 2.7 MB.
Code & Tests
Datasets
People & History
- 0.1.1Latest2026-10-20 · current release ·
12 days in CRAN's queue
Listed in CRAN's incoming queue from 2026-10-08 20:25 to 2026-10-20 12:27 UTC, in 62 readings. It was also listed earlier, first on 2026-10-03 21:57 UTC, and left the queue for more than a day before this stay. - RR 4.6.0 released · 2026-04-24
Package metadata
- First published
- 2026-10-20
- Total releases
- 1 / 1 yrs
- License
- LGPL-3 OSI
- Minimum R
- ≥ 4.1
- Bundled data
- 403 KB / 2 files
- Download size
- 2.4 MB
Cite
Cite this package
Run in R for the authors' preferred citation:
citation("DistributionalDDA")This package has no citation file, so citation() returns this entry built from DESCRIPTION.
Cite the R Observatory
For a number measured here: a download total, a coverage figure, an archival date.
From data release v2026-10-21, which the citation names so these numbers can be found later. More on citing and the projects behind them.