Premium
Robust partial least squares regression: Part II, new algorithm and benchmark studies
Author(s) -
Kruger Uwe,
Zhou Yan,
Wang Xun,
Rooney David,
Thompson Jillian
Publication year - 2008
Publication title -
journal of chemometrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.47
H-Index - 92
eISSN - 1099-128X
pISSN - 0886-9383
DOI - 10.1002/cem.1095
Subject(s) - partial least squares regression , outlier , benchmark (surveying) , algorithm , sensitivity (control systems) , estimator , computer science , regression , robust regression , work (physics) , data mining , mathematics , statistics , artificial intelligence , machine learning , engineering , geodesy , electronic engineering , geography , mechanical engineering
This paper presents the second part of the work on robust partial least squares (RPLS) regression and develops a new RPLS algorithm based on the concept laid out in Part I. The paper also contrasts the new algorithm with existing work using two simulation examples. This comparison highlights (i) the impact of the flaws in existing RPLS work and (ii) the compromised sensitivity resulting from introducing simplifications to the determination of the Stahel–Donoho estimator (SDE). The paper finally presents an evaluation of the computational complexity of RPLS algorithms and examines the impact of the signal‐to‐noise ratio (SNR) upon the sensitivity of detecting outliers. The third part of this work will examine practical aspects of RPLS applications based on the analysis of experimental data. Copyright © 2007 John Wiley & Sons, Ltd.