Evaluation and comparison of different approaches to multi-product brix calibration in near-infrared spectroscopy
Near-infrared (NIR) spectroscopy became a widespread technology for qualitative and quantitative material analysis. New fields of application of this technology, e.g., quantitative food analysis for consumers, increase demand for multiproduct calibration models. Conventional multivariate calibration methods, such as partial least squares regression (PLSR), are reported to show weakness in predictive performance . Preliminary studies in multi-product calibration for quantitative analysis of food with near-infrared spectroscopy showed good results for memory-based learning (MBL) and a classification prediction hierarchy (CPH) . In this study, three varieties of apples, pears and tomatoes with known °brix value are analyzed with NIR spectroscopy in the range from 900 nm to 2400 nm. Predictive performance of a linear PLSR model, two nonlinear models (CPH and MBL) and different preprocessing techniques are tested and evaluated. For error estimation, leave-oneproduct-out and leave-one-out cross-validation are used.