Figure 6.
Result for the contiguous-block, cross-validation method performed on the covariance matrix constructed of 10 fluorescence spectra. RMSECV curve normally has a break or minimum against the true number of significant factors (Malinowski 2002; Wise et al. 2005). Factors which decrease RMSECV by <2% of the maximum RMSECV are usually rejected as error factors (Wise et al. 2005).