Wedding the wavelet transform and multivariate data analysis

Fionn Murtagh

Research output: Contribution to journalArticlepeer-review

10 Citations (Scopus)


We discuss the use of orthogonal wavelet transforms in preprocessing multivariate data for subsequent analysis, e.g., by clustering or dimensionality reduction. Wavelet transforms allow us to introduce multiresolution approximation, and multiscale nonparametric regression or smoothing, in a natural and integrated way into the data analysis. As will be explained in the first part of the paper, this approach is of greatest interest for multivariate data analysis when we use (i) datasets with ordered variables, e.g., time series, and (ii) object dimensionalities which are not too small, e.g., 16 and upwards. In the second part of the paper, a different type of wavelet decomposition is used. Applications illustrate the powerfulness of this new perspective on data analysis.

Original languageEnglish
Pages (from-to)161-183
Number of pages23
JournalJournal of Classification
Issue number2
Publication statusPublished - 1 Feb 1998
Externally publishedYes


Dive into the research topics of 'Wedding the wavelet transform and multivariate data analysis'. Together they form a unique fingerprint.

Cite this