
Open access
Author
Date
2018-05Type
- Journal Article
Abstract
The estimation of probability densities based on available data is a central task in many statistical applications. Especially in the case of large ensembles with many samples or high-dimensional sample spaces, computationally efficient methods are needed. We propose a new method that is based on a decomposition of the unknown distribution in terms of so-called distribution elements (DEs). These elements enable an adaptive and hierarchical discretization of the sample space with small or large elements in regions with smoothly or highly variable densities, respectively. The novel refinement strategy that we propose is based on statistical goodness-of-fit and pairwise (as an approximation to mutual) independence tests that evaluate the local approximation of the distribution in terms of DEs. The capabilities of our new method are inspected based on several examples of different dimensionality and successfully compared with other state-of-the-art density estimators. Show more
Permanent link
https://doi.org/10.3929/ethz-b-000168068Publication status
publishedExternal links
Journal / series
Statistics and ComputingVolume
Pages / Article No.
Publisher
SpringerSubject
Nonparametric density estimation; Adaptive histogram; Kernel density estimation; Adaptive binning; Polynomial histogram; Curse of dimensionality; High dimensional; Big data; Pólya tree; Density estimation treeOrganisational unit
03644 - Jenny, Patrick / Jenny, Patrick
Notes
It was possible to publish this article open access thanks to a Swiss National Licence with the publisher.More
Show all metadata