Numerical Considerations and a New Implementation for Invariant Coordinate Selection
Aurore Archimbaud,
Zlatko Drmac,
Klaus Nordhausen,
Una Radojicic and
Anne Ruiz-Gazen
Additional contact information
Aurore Archimbaud: Erasmus University Rotterdam
Zlatko Drmac: University of Zagreb
Klaus Nordhausen: JYU - University of Jyväskylä
Una Radojicic: TU Wien - Vienna University of Technology = Technische Universität Wien
Post-Print from HAL
Abstract:
Invariant coordinate selection (ICS) is a multivariate data transformation and a dimension reduction method that can be useful in many different contexts. It can be used for outlier detection or cluster identification, and can be seen as an independent component or a non-Gaussian component analysis method. The usual implementation of ICS is based on a joint diagonalization of two scatter matrices, and may be numerically unstable in some ill-conditioned situations. We focus on one-step M-scatter matrices and propose a new implementation of ICS based on a pivoted QR factorization of the centered data set. This factorization avoids the direct computation of the scatter matrices and their inverse and brings numerical stability to the algorithm. Furthermore, the row and column pivoting leads to a rank revealing procedure that allows computation of ICS when the scatter matrices are not full rank. Several artificial and real data sets illustrate the interest of using the new implementation compared to the original one.
Keywords: Dimension reduction; Invariant coordinate selection; One-step M-estimators; Pivoting; QR factorization; Scatter matrices (search for similar items in EconPapers)
Date: 2023
References: Add references at CitEc
Citations: View citations in EconPapers (1)
Published in SIAM Journal on Mathematics of Data Science, 2023, 5 (1), pp.97-121. ⟨10.1137/22M1498759⟩
There are no downloads for this item, see the EconPapers FAQ for hints about obtaining it.
Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.
Export reference: BibTeX
RIS (EndNote, ProCite, RefMan)
HTML/Text
Persistent link: https://EconPapers.repec.org/RePEc:hal:journl:hal-04038657
DOI: 10.1137/22M1498759
Access Statistics for this paper
More papers in Post-Print from HAL
Bibliographic data for series maintained by CCSD ().