Cd. Brown et al., Derivative preprocessing and optimal corrections for baseline drift in multivariate calibration, APPL SPECTR, 54(7), 2000, pp. 1055-1068
The characteristics of baseline drift are discussed from the perspective of
error covariance. From this standpoint, the operation of derivative filter
s as preprocessing tools for multivariate calibration is explored. It is sh
own that convolution of derivative filter coefficients with the error covar
iance matrices for the data tend to reduce the contributions of correlated
error, thereby reducing the presence of drift noise. This theory is corrobo
rated by examination of experimental error covariance matrices before and a
fter derivative preprocessing. It is proposed that maximum likelihood princ
ipal components analysis (MLPCA) is an optimal method for countering the de
leterious effects of drift noise when the characteristics of that noise are
known, since MLPCA uses error covariance information to perform a maximum
likelihood projection of the data. In simulation and experimental studies,
the performance of MLPCR and derivative-preprocessed PCR are compared to th
at of PCR with multivariate calibration data showing significant levels of
drift. MLPCR is found to perform as well as or better than derivative PCR (
with the best-suited derivative filter characteristics), provided that reas
onable estimates of the drift noise characteristics are available. Recommen
dations are given for the use of MLPCR with poor estimates of the error cov
ariance information.