Block-bootstrapping for noisy data

Malenka Mader*, Wolfgang Mader, Linda Sommerlade, Jens Timmer, Bjoern Schelter

*Corresponding author for this work

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

Background: Statistical inference of signals is key to understand fundamental processes in the neurosciences. It is essential to distinguish true from random effects. To this end, statistical concepts of confidence intervals, significance levels and hypothesis tests are employed. Bootstrap-based approaches complement the analytical approaches, replacing the latter whenever these are not possible.

New method: Block-bootstrap was introduced as an adaption of the ordinary bootstrap for serially correlated data. For block-bootstrap, the signals are cut into independent blocks, yielding independent samples. The key parameter for block-bootstrapping is the block length. In the presence of noise, naive approaches to block-bootstrapping fail. Here, we present an approach based on block-bootstrapping which can cope even with high noise levels. This method naturally leads to an algorithm of block-bootstrapping that is immediately applicable to observed signals.

Results: While naive block-bootstrapping easily results in a misestimation of the block length, and therefore in an over-estimation of the confidence bounds by 50%, our new approach provides an optimal determination of these, still keeping the coverage correct.

Comparison with existing methods: In several applications bootstrapping replaces analytical statistics. Block-bootstrapping is applied to serially correlated signals. Noise, ubiquitous in the neurosciences, is typically neglected. Our new approach not only explicitly includes the presence of (observational) noise in the statistics but also outperforms conventional methods and reduces the number of false-positive conclusions.

Conclusions: The presence of noise has impacts on statistical inference. Our ready-to-apply method enables a rigorous statistical assessment based on block-bootstrapping for noisy serially correlated data.

Original languageEnglish
Pages (from-to)285-291
Number of pages7
JournalJournal of Neuroscience Methods
Volume219
Issue number2
Early online date8 Aug 2013
DOIs
Publication statusPublished - 15 Oct 2013

Keywords

  • Statistics
  • Distribution estimation
  • Autocorrelation
  • Measurement noise
  • Dependent data
  • Tremor

Cite this

Block-bootstrapping for noisy data. / Mader, Malenka; Mader, Wolfgang; Sommerlade, Linda; Timmer, Jens; Schelter, Bjoern.

In: Journal of Neuroscience Methods, Vol. 219, No. 2, 15.10.2013, p. 285-291.

Research output: Contribution to journalArticle

Mader, M, Mader, W, Sommerlade, L, Timmer, J & Schelter, B 2013, 'Block-bootstrapping for noisy data', Journal of Neuroscience Methods, vol. 219, no. 2, pp. 285-291. https://doi.org/10.1016/j.jneumeth.2013.07.022
Mader, Malenka ; Mader, Wolfgang ; Sommerlade, Linda ; Timmer, Jens ; Schelter, Bjoern. / Block-bootstrapping for noisy data. In: Journal of Neuroscience Methods. 2013 ; Vol. 219, No. 2. pp. 285-291.
@article{a6065fc1b61e48bda923c718124b06d3,
title = "Block-bootstrapping for noisy data",
abstract = "Background: Statistical inference of signals is key to understand fundamental processes in the neurosciences. It is essential to distinguish true from random effects. To this end, statistical concepts of confidence intervals, significance levels and hypothesis tests are employed. Bootstrap-based approaches complement the analytical approaches, replacing the latter whenever these are not possible.New method: Block-bootstrap was introduced as an adaption of the ordinary bootstrap for serially correlated data. For block-bootstrap, the signals are cut into independent blocks, yielding independent samples. The key parameter for block-bootstrapping is the block length. In the presence of noise, naive approaches to block-bootstrapping fail. Here, we present an approach based on block-bootstrapping which can cope even with high noise levels. This method naturally leads to an algorithm of block-bootstrapping that is immediately applicable to observed signals.Results: While naive block-bootstrapping easily results in a misestimation of the block length, and therefore in an over-estimation of the confidence bounds by 50{\%}, our new approach provides an optimal determination of these, still keeping the coverage correct.Comparison with existing methods: In several applications bootstrapping replaces analytical statistics. Block-bootstrapping is applied to serially correlated signals. Noise, ubiquitous in the neurosciences, is typically neglected. Our new approach not only explicitly includes the presence of (observational) noise in the statistics but also outperforms conventional methods and reduces the number of false-positive conclusions.Conclusions: The presence of noise has impacts on statistical inference. Our ready-to-apply method enables a rigorous statistical assessment based on block-bootstrapping for noisy serially correlated data.",
keywords = "Statistics, Distribution estimation, Autocorrelation, Measurement noise, Dependent data, Tremor",
author = "Malenka Mader and Wolfgang Mader and Linda Sommerlade and Jens Timmer and Bjoern Schelter",
note = "We thank Bernhardt Hellwig and Florian Amtage for the tremor data and the fruitful discussions.",
year = "2013",
month = "10",
day = "15",
doi = "10.1016/j.jneumeth.2013.07.022",
language = "English",
volume = "219",
pages = "285--291",
journal = "Journal of Neuroscience Methods",
issn = "0165-0270",
publisher = "Elsevier",
number = "2",

}

TY - JOUR

T1 - Block-bootstrapping for noisy data

AU - Mader, Malenka

AU - Mader, Wolfgang

AU - Sommerlade, Linda

AU - Timmer, Jens

AU - Schelter, Bjoern

N1 - We thank Bernhardt Hellwig and Florian Amtage for the tremor data and the fruitful discussions.

PY - 2013/10/15

Y1 - 2013/10/15

N2 - Background: Statistical inference of signals is key to understand fundamental processes in the neurosciences. It is essential to distinguish true from random effects. To this end, statistical concepts of confidence intervals, significance levels and hypothesis tests are employed. Bootstrap-based approaches complement the analytical approaches, replacing the latter whenever these are not possible.New method: Block-bootstrap was introduced as an adaption of the ordinary bootstrap for serially correlated data. For block-bootstrap, the signals are cut into independent blocks, yielding independent samples. The key parameter for block-bootstrapping is the block length. In the presence of noise, naive approaches to block-bootstrapping fail. Here, we present an approach based on block-bootstrapping which can cope even with high noise levels. This method naturally leads to an algorithm of block-bootstrapping that is immediately applicable to observed signals.Results: While naive block-bootstrapping easily results in a misestimation of the block length, and therefore in an over-estimation of the confidence bounds by 50%, our new approach provides an optimal determination of these, still keeping the coverage correct.Comparison with existing methods: In several applications bootstrapping replaces analytical statistics. Block-bootstrapping is applied to serially correlated signals. Noise, ubiquitous in the neurosciences, is typically neglected. Our new approach not only explicitly includes the presence of (observational) noise in the statistics but also outperforms conventional methods and reduces the number of false-positive conclusions.Conclusions: The presence of noise has impacts on statistical inference. Our ready-to-apply method enables a rigorous statistical assessment based on block-bootstrapping for noisy serially correlated data.

AB - Background: Statistical inference of signals is key to understand fundamental processes in the neurosciences. It is essential to distinguish true from random effects. To this end, statistical concepts of confidence intervals, significance levels and hypothesis tests are employed. Bootstrap-based approaches complement the analytical approaches, replacing the latter whenever these are not possible.New method: Block-bootstrap was introduced as an adaption of the ordinary bootstrap for serially correlated data. For block-bootstrap, the signals are cut into independent blocks, yielding independent samples. The key parameter for block-bootstrapping is the block length. In the presence of noise, naive approaches to block-bootstrapping fail. Here, we present an approach based on block-bootstrapping which can cope even with high noise levels. This method naturally leads to an algorithm of block-bootstrapping that is immediately applicable to observed signals.Results: While naive block-bootstrapping easily results in a misestimation of the block length, and therefore in an over-estimation of the confidence bounds by 50%, our new approach provides an optimal determination of these, still keeping the coverage correct.Comparison with existing methods: In several applications bootstrapping replaces analytical statistics. Block-bootstrapping is applied to serially correlated signals. Noise, ubiquitous in the neurosciences, is typically neglected. Our new approach not only explicitly includes the presence of (observational) noise in the statistics but also outperforms conventional methods and reduces the number of false-positive conclusions.Conclusions: The presence of noise has impacts on statistical inference. Our ready-to-apply method enables a rigorous statistical assessment based on block-bootstrapping for noisy serially correlated data.

KW - Statistics

KW - Distribution estimation

KW - Autocorrelation

KW - Measurement noise

KW - Dependent data

KW - Tremor

U2 - 10.1016/j.jneumeth.2013.07.022

DO - 10.1016/j.jneumeth.2013.07.022

M3 - Article

VL - 219

SP - 285

EP - 291

JO - Journal of Neuroscience Methods

JF - Journal of Neuroscience Methods

SN - 0165-0270

IS - 2

ER -