Classification of “Ricotta” whey cheese from different milk and Designation of Origin-protected samples through infrared spectroscopy and chemometric analysis

Whey cheeses are produced in various parts of the world, such as Portugal, Spain, and Turkey. In Italy, whey cheese goes under the name “ricotta”. This study investigates the classification of ricotta whey cheese derived from various milk sources (either protected designation of origin (PDO) or not) using an Attenuated Total Reflectance Fourier-transform infrared (ATR-FTIR) spectroscopy combined with chemometric analysis. Employing the SPORT-LDA method, which can incorporate Variable Importance in Projection (VIP) analysis, 287 samples of ricotta cheese produced using milk from four different animals (sheep, cow, goat, and water buffalo) were classified according to the animal origin. This led to the correct classification of 97 % of the test samples (3 misclassified samples over 97). VIP analysis revealed that the spectral ranges of 3300–3100 cm⁻¹, 2900–2800 cm⁻¹, and 1700–1300 cm⁻¹ are consistently relevant across all milk sources, thanks to the key molecular vibrations associated with protein structures, lipid content, and water. Eventually, the analysis was circumscribed to sheep ricotta cheeses, because some of these present the PDO quality mark. SIMCA was used to classify PDO samples with respect to the Non-PDO sheep ricotta individuals. The application of SIMCA to model class PDO led to 82.1 % of sensitivity and 82.7 % of specificity (in external validation). The findings underscore the robustness of ATR-FTIR spectroscopy and chemometrics in maintaining the integrity of PDO products and ensuring quality control. • Ricotta whey samples are analyzed by FT-IR spectroscopy and chemometrics. • Ricotta samples are discriminated according to animal origin of milk by SPORT-LDA. • Using SPORT-LDA to discriminate, the correct classification rate (on test set) is 97 %. • Sheep ricotta samples PDO are modeled by SIMCA. • SIMCA led to 98.2 % specificity and 91.7 % sensitivity, on test samples.