Premium
Response to ‘Comments on construction of confidence intervals in connection with partial least squares’
Author(s) -
Faber Nicolaas Klaas M.
Publication year - 2000
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 - 10.1002/1099-128x(200007/08)14:4<363::aid-cem580>3.0.co;2-7
Subject(s) - linearization , partial least squares regression , chemometrics , mathematics , statistics , least squares function approximation , confidence interval , econometrics , computer science , nonlinear system , physics , machine learning , quantum mechanics , estimator
Recently, Denham, Phatak and co‐workers, and Faber and Kowalski have proposed local linearization to obtain prediction intervals when partial least squares has been used as estimation method. This approach has been criticized by Morsing and Ekman to ignore bias in the regression coefficient estimates ( J. Chemometrics 1998; 12: 295–299). This response shows that, in contrast with Morsing and Ekman's claims, considerable attention has been paid to bias in the original work. In addition, it is explained that in typical chemometrics applications (inverse calibration of, for example, near‐infrared spectroscopic data) bias is likely to be small, which implies that the original local linearization approach is potentially useful in practice. Finally, a method is proposed to include non‐negligible bias in the regression coefficient estimates in the prediction intervals. Copyright © 2000 John Wiley & Sons, Ltd.