COMPARISON OF SAMPLE HANDLING AND DATA TREATMENT METHODS FOR DETERMINING MOISTURE AND FAT IN CHEDDAR CHEESE BY NEAR-INFRARED SPECTROSCOPY

Various sample handling and data treatment methods were compared for determining moisture and fat in Cheddar cheese by near infrared (near-IR) spectroscopy. Reflectance and transmittance spectra were collected from grated samples, and reflectance spectra were also collected from intact samples with a fiber optic probe. Calibrations were developed using partial least-squares (PLS) and multiple Linear regression (MLR) techniques. For grated samples, PLS regression of first-derivative reflectance data yielded a calibration that gave the lowest standard error of prediction (SEP) for moisture (0.335%), while a PLS regression of log 1/R spectral data performed best for determining fat (SEP = 0.330%). Acceptable results for moisture were also obtained from a calibration based on MLR of first-derivative transmittance spectra (SEP = 0.356%), but determining fat by transmittance yielded high SEP values. Use of a fiber optic probe with intact cheese blocks gave less reliable results than were obtained from grated samples.