APPLICATION OF MULTIPLE VARIABLE ANALYSIS TO ABSORPTION SPECTROSCOPY OF ACTINIDES

Year
1989
Author(s)
Patrick E. O'Rourke - Savannah River Laboratory
Abstract
A multivariate analysis program (SRLMVA) was developed to predict chemical concentrations from a solution's absorption spectrum. Multivariate calibration models are more accurate, reliable and robust than conventional univariate calibrations but are more difficult to build and verify. SRLMVA provides users with most of the functions necessary to explore correlations between data and concentrations, process data, build multivariate calibration models, and test the models. Model types include multiple linear regression (MLR), classical least squares curve resolution (CLS), principle component regression (PCR) and partial least squares regression (PLS). Data processing functions include derivatives, fast Fourier transforms, background subtraction, and data scaling functions. The program as developed is quite general and can be applied to any spectral or chromatographic data.