Determination of nutrient composition of some compound feeds and feeding stuffs by FTIR- spectroscopy combined with partial least sguare regression (PLSR)

Fourier Transform Infrared (FTIR) Spectroscopic measurement has become one of the most widely used analysis techniques due to the being a fast, reliable, sensitive and inexpensive technique that can be used to analyze the chemical components and even the molecular structure of various microorganisms, food and feed, and different organic and inorganic compounds. The present study, five grains (bran, wheat, razmol, barley, and oats), three meals (soybean, sunflower and cotton) and five mixed feed (broiler chicken, laying chicken, bovine milk, and cattle fattening feed) were analyzed for their chemical contents. A mathematical model was created using Partial Least Squares Regression (PLSR) method in order to estimate the nutritional composition of feeds from FTIR spectrum data. The model 1 performance characteristic were r2:0,867665, R2:0,803671, RMSECV:0,2296 respectively. In addition, the performance charecteristics of model 2 were found as r2:0,80, R2:0,73, RMSECV:0,3465 respectively. Determined by both methods; In terms of dry matter (KM), crude protein (HP), crude ash (HK), the results were very close to each other and a high level of correlation was found. Herewith the study, the results revealed that the nutrient contents of the feed and mixed feedstuff can be estimated using FTIR measurements. Based on these results, the method is recommended to apply to the routine analysis in food and feed industry. Keywords: Feedstuff, Nutrient, Fourier Transform Infrared (FTIR) Spectroscopic Measurement, Partial Least Squares Regression