Submitted to: Meeting Abstract
Publication Type: Abstract Only
Publication Acceptance Date: August 18, 2005
Publication Date: November 13, 2005
Citation: Reeves III, J.B., Delwiche, S.R., Palmquist, D.E., Reeves, V.B. 2005. Least-squares means multiple comparison testing of actual versus predicted residuals for evaluation of partial least squares (pls) spectral calibrations [abstract]. Technical Abstract: It is common to use data pre-treatments such as scatter correction, derivatives, mean centering, variance scaling, etc. prior to the development of near- and mid-infrared spectral calibrations. As a result, it is possible to generate a multitude of calibrations many of which will have similar statistical properties, e.g. similar r-square or error values. Because calibrations always fit the data upon which they were developed best, the absolute best calibration out of a number may not be the best for determining the values of future samples, e.g. not the most robust calibration. If several calibrations are found to be statistical the same, then other criteria could be used to determine which one to use (e.g., one with fewest PLS factors or based on past experience) or further investigations could be carried out on the more limited set of calibrations deemed to represent the best of all those originally developed. However, there is no accepted statistical procedure for determining which calibrations are statistically the same and which are not from a large group of calibrations. We propose the use of least-squares means multiple comparison testing of actual versus predicted residuals for determining the statistical similarity of multiple PLS calibrations. A program has been developed using SAS Proc Mixed which computes and summarizes comparisons of PLS calibrations. This method would also be applicable to other calibration methods (e.g., PCR, MLR, etc.) as it only requires a list of actual and predicted values for each calibration as input.