SIMPLE MULTIVARIATE CALIBRATION METHOD WITH AN APPROPRIATE NUMBER OF PRINCIPAL COMPONENTS USING SINGULAR-VALUE DECOMPOSITION AND CROSS-VALIDATION PROCEDURE

Based on a simple model of multivariate calibration, a cross-validation procedure was performed for estimating the number of principal components with the help of singular value decomposition (SVD). The multivariate calibration method with an appropriate number of principal components has been used to determine simultaneously several components in composite aspirin tablets and in composite vitamin B tablets with satisfactory precision and accuracy. The result obtained by the method is very similar to that by partial least squares (PLS) and superior to that by multiple linear regression (MLR). Provided that the matrix composition of the calibration system is similar to that of the analysis system, the method is suitable for the direct or indirect calibration of multicomponent systems or of real samples with incomplete information.