wbacon

Weighted BACON algorithms for multivariate outlier nomination (detection) and robust linear regression

CRAN CRAN

Summary

Outlier nomination (detection) and robust regression are computationally hard problems. This is all the more true when the number of variables and observations grow rapidly. Among all candidate methods, the two BACON (blocked adaptive computationally efficient outlier nominators) algorithms of Billor et al. (2000) have favorable computational characteristics as they require only a few model evaluations irrespective of the sample size. This makes them popular algorithms for multivariate outlier nomination/ detection and robust linear regression (in early 2021, Google Scholar reports more than 500 citations of the Billor et al. (2000) paper).

wbacon is a package for the R statistical software. It is aimed at medium to large data sets that can possibly have (sampling) weights (e.g., data from complex survey samples). The package has a user-friendly R interface (with plotting methods, etc.) and is written mainly in the C language (with OpenMP support for parallelization for performance reasons.


Methods

Available methods

  • wBACON() is for multivariate outlier nomination and robust estimation of location/ center and covariance matrix
  • wBACON_reg() is for robust linear regression (the method is robust against outliers in the response variable and the model's design matrix)
and diagnostic tools; see vignette below.


Documentation

The paper Schoch (2021) gives an overview. Further details can be found in:
  • Vignette, link
  • Technical documentation, link


Code respository


GitHub
https://github.com/tobiasschoch/wbacon

CRAN
https://CRAN.R-project.org/package=wbacon


References

  • BILLOR, N., HADI, A. S., and VELLEMAN , P. F. (2000). BACON: Blocked adaptive computationally-efficient outlier nominators, Computational Statistics and Data Analysis 34, 279-298. DOI 10.1016/S0167-9473(99)00101-2
  • SCHOCH, T. (2021). wbacon: Weighted BACON algorithms for multivariate outlier nomination (detection) and robust linear regression, Journal of Open Source Software 6 (62), 3238. DOI 10.21105/joss.03238