Non-destructive/Non-invasive Method Development, Evaluation, and Transfer

This chapter attempts to give an overview of the main aspects to be taken into account for spectral data processing using chemometrics, considering the development and validation of calibrations and qualitative predictive models, considerations on signal pretreatments, possibilities for instrument standardization and model transfer, together with the final performance of predictive models for routine use. The main focus will be on NIR spectroscopy. However, the general principles discussed in the chapter are transferable to other non-invasive methodologies that produce multivariate raw measurements. The authors wish to emphasize those aspects that, in their experience, are critical in the development of applications using data from nondestructive spectral sensors.