Exact fast computation of band depth for large functional datasets: How quickly can one million curves be ranked?

Ying Sun*, Marc G. Genton, Douglas W. Nychka

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

35 Scopus citations

Abstract

Band depth is an important nonparametric measure that generalizes order statistics and makes univariate methods based on order statistics possible for functional data. However, the computational burden of band depth limits its applicability when large functional or image datasets are considered. This paper proposes an exact fast method to speed up the band depth computation when bands are defined by two curves. Remarkable computational gains are demonstrated through simulation studies comparing our proposal with the original computation and one existing approximate method. For example, we report an experiment where our method can rank one million curves, evaluated at fifty time points each, in 12.4 seconds with Matlab.

Original languageEnglish (US)
Pages (from-to)68-74
Number of pages7
JournalStat
Volume1
Issue number1
DOIs
StatePublished - Oct 2012

Keywords

  • Approximate solution
  • Band depth
  • Exact solution
  • Functional boxplot
  • Functional data
  • Large dataset
  • Modified band depth

ASJC Scopus subject areas

  • Statistics and Probability
  • Statistics, Probability and Uncertainty

Fingerprint

Dive into the research topics of 'Exact fast computation of band depth for large functional datasets: How quickly can one million curves be ranked?'. Together they form a unique fingerprint.

Cite this