Benchmark testing of algorithms for very robust regression: FS, LMS and LTS

@article{Torti2012BenchmarkTO,
  title={Benchmark testing of algorithms for very robust regression: FS, LMS and LTS},
  author={Francesca Torti and Domenico Perrotta and Anthony C. Atkinson and Marco Riani},
  journal={Computational Statistics & Data Analysis},
  year={2012},
  volume={56},
  pages={2501-2512}
}
The methods of very robust regression resist up to 50% of outl iers. The algorithms for very robust regression rely on selecting numerous subsa mples of the data. We describe new algorithms for LMS and LTS estimators that have in creased efficiency due to improved combinatorial sampling. These and other public ly available algorithms are compared for outlier detection. An algorithm using the f orward search has the best properties for both size and power of the outlier tests. 

From This Paper

Figures, tables, and topics from this paper.

References

Publications referenced by this paper.
Showing 1-10 of 14 references

Plots, Transformations, and Regression

  • A. C. Atkinson
  • Computational Statistics and Data Analysis
  • 1985
Highly Influential
3 Excerpts

Beyond location parameters: robust concepts and methods

  • F. Torti
  • Computational Statistics and Data Analysis
  • 2012
1 Excerpt

Advances in the Forward Search: Methodological and Applied Contributions. Italian Statistical Society, Padova

  • F. Torti
  • Chemometrics and Intelligent Laboratory Systems
  • 2011
1 Excerpt

Matlab library LIBRA

  • M. Hubert
  • WIREs Computational Statistics
  • 2010
1 Excerpt

Robust Statistics: Theory and Methods

  • D. R. Martin, V. J. Yohai
  • 2006