Anal. Chem. 2000, 72, 1944
Correction A Transparent Tool for Seemingly Difficult Calibrations: The Parallel Calibration Method Clifford H. Spiegelman, Jerome F. Bennett, Marina Vannucci, Michael J. McShane, and Gerald L. Cote (Anal. Chem. 2000, 72, 135-140).
We would like to thank Dr. Paul Kramer and Richard Spragg for pointing out to us an earlier presentation using the algorithm in our paper, which was published by G. L. McClure, P. B. Roush, J. F. Williams, and C. Lehmann, “Application of Computerized Infrared Spectroscopy to the Analysis of the Principal Lipids Found in Blood Serum,” in: Computerized Quantitative Infrared Analysis, pps. 131-179, ASTM Special Technical Publication 934, G. L. McClure, Ed. American Society of Testing Materials, Philadelphia, PA, 1987. Additionally, Tables 1 and 2 were printed incorrectly. The correct tables are shown below. Table 1. Mean Squared Errors for Pseudogasoline Dataa method
G1
G2
G3
G4
G5
aveb
corrc
PLS PAR CLS
0.39 0.12 13.95
0.11 0.07 39.96
0.09 0.08 9.78
0.34 0.10 74.86
0.21 0.07 25.19
0.23 0.09 32.75
1.00 1.00 0.59
a Mean squared errors were computed for test sets comprising 20% of the total sample sizes b The average MSE over the ingredients for each method. c The correlation between the predicted concentrations and the true concentration for each method.
Table 2. Mean Squared Errors for Aqueous Mixture Dataa method
glucose
lactate
ammonia
glutamate
glutamine
aveb
corrc
PLS PAR CLS
0.41 0.12 3.62
0.19 0.01 0.13
0.08 0.00 0.09
0.26 0.01 0.23
0.33 0.12 2.70
0.25 0.05 1.36
0.99 1.00 0.59
a Leave-one-out cross-validation was used to estimate mean squared errors. b The average MSE over the ingredients for each method. c The correlation between the predicted concentrations and the true concentration for each method.
AC001704U 10.1021/ac001704u Published on Web 03/21/2000