EconStor >
Universität zu Köln >
Institut für Ökonometrie und Statistik, Universität Köln >
Discussion Papers in Econometrics and Statistics, Institut für Ökonometrie und Statistik, Universität Köln >

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10419/45356
  
Title:An exact algorithm for weighted-mean trimmed regions in any dimension PDF Logo
Authors:Bazovkin, Pavel
Mosler, Karl
Issue Date:2010
Series/Report no.:Discussion papers in statistics and econometrics 6/10
Abstract:Trimmed regions are a powerful tool of multivariate data analysis. They describe a probability distribution in Euclidean d-space regarding location, dispersion, and shape, and they order multivariate data with respect to their centrality. Dyckerhoff and Mosler (201x) have introduced the class of weighted-mean trimmed regions, which possess attractive properties regarding continuity, subadditivity, and monotonicity. We present an exact algorithm to compute the weighted-mean trimmed regions of a given data cloud in arbitrary dimension d. These trimmed regions are convex polytopes in Rd. To calculate them, the algorithm builds on methods from computational geometry. A characterization of a region's facets is used, and information about the adjacency of the facets is extracted from the data. A key problem consists in ordering the facets. It is solved by the introduction of a tree-based order. The algorithm has been programmed in C++ and is available as an R package.
Subjects:central regions
data depth
multivariate data analysis
convex polytope
computational geometry
algorithm
C++, R
Document Type:Working Paper
Appears in Collections:Discussion Papers in Econometrics and Statistics, Institut für Ökonometrie und Statistik, Universität Köln

Files in This Item:
File Description SizeFormat
656640324.pdf2.33 MBAdobe PDF
No. of Downloads: Counter Stats
Download bibliographical data as: BibTeX
Share on:http://hdl.handle.net/10419/45356

Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.