Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Algorithms for robust nonlinear regression with heteroscedastic errors

L Tóthfalusi1, L Endrényi

  • 1Department of Pharmacology, Department of Preventive Medicine and Biostatistics, University of Toronto, Toronto, Ontario, Canada.

International Journal of Bio-Medical Computing
|August 1, 1996
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Effect of a single neonatal oxytocin treatment (hormonal imprinting) on the biogenic amine level of the adult rat brain: could oxytocin-induced labor cause pervasive developmental diseases?

Reproductive sciences (Thousand Oaks, Calif.)·2013
Same author

Estimating product bioequivalence for highly variable veterinary drugs.

Journal of veterinary pharmacology and therapeutics·2012
Same author

Bioequivalence: tried and tested.

Cardiovascular journal of Africa·2010
Same author

Effect of neonatal benzpyrene imprinting on the brain serotonin content and nocistatin level in adult male rats.

Acta physiologica Hungarica·2007
Same author

Effects of extremely low concentrations of hormones on the insulin binding of Tetrahymena.

Cell biology international·2006
Same author

Prolonged effect of stress (water and food deprivation) at weaning or in adult age on the triiodothyronine and histamine content of immune cells.

Hormone and metabolic research = Hormon- und Stoffwechselforschung = Hormones et metabolisme·2005

This study compared nonlinear regression algorithms using Monte-Carlo simulations, focusing on heteroscedasticity and outliers. Extended least squares (ELSQ) proved most reliable for parameter estimation efficiency and outlier resistance.

Area of Science:

  • Statistical modeling
  • Computational statistics

Background:

  • Nonlinear regression is crucial for analyzing complex data.
  • Measurement errors can be dependent on values (heteroscedasticity) and include outliers, complicating analysis.
  • Standard least squares (LSQ) methods may fail under these conditions.

Purpose of the Study:

  • To compare the performance of various nonlinear regression algorithms.
  • To evaluate algorithms under conditions of heteroscedasticity and outliers.
  • To identify the most reliable algorithm for parameter estimation.

Main Methods:

  • Monte Carlo simulations were employed to generate data with heteroscedasticity and outliers.
  • Least squares (LSQ) algorithms, including those with user-supplied or estimated weights (e.g., extended least squares - ELSQ), were tested.

Related Experiment Videos

  • Robust regression methods, such as robust iteratively reweighted (IRR) and least absolute value (LAV) regressions, were also evaluated.

    • Main Results:

    • Extended least squares (ELSQ) demonstrated the highest reliability among tested LSQ algorithms.
    • Robust iteratively reweighted (IRR) algorithms showed slightly higher efficiency than least absolute value (LAV) algorithms in the absence of outliers.
    • Least absolute value (LAV) algorithms offered superior protection against outliers compared to IRR algorithms.

    Conclusions:

    • Extended least squares (ELSQ) is recommended for nonlinear regression when dealing with heteroscedasticity and potential outliers.
    • The choice between IRR and LAV robust methods depends on the expected presence of outliers.
    • Robust methods are essential for reliable parameter estimation in the presence of data contamination.