Modelo de predicción para las variables físico químicas que generan devolución del producto leche semidescremada UHT polietileno 1L, a partir del uso de técnicas de regresión logística para el área de calidad en el sector lácteo

Abstract

The rejection of a product by consumers generates financial consequences and an increase of nonconforming products within a company dedicated to the production of dairy products; that is why, in order to comply with the food safety conditions of the product, adequate quality controls must be established. This research was based on identifying the product with the highest number of returns made by consumers to the Quality Management Area of the company Lácteos San Antonio, being the Semi-skimmed Milk UHT polyethylene 1L. The physicochemical variables were studied in order to propose a prediction model that would allow reducing the amount of nonconforming product. To this end, a deductive and correlational research methodology was applied, starting from the analysis of the returned products with the most representative conditions for rejection, being the polyethylene sleeves cut, swollen and dented, with 43%, 20% and 14%, respectively. An ANOVA analysis was performed for the pH, acidity and cryoscopy variables, which determined that cryoscopy had no influence on the development of the prediction model. In addition, logistic regression analysis led to the determination that pH alone is the predictor variable. Finally, from the prediction model, the lower pH range was adjusted from 6.55 to 6.60 and a decrease in the percentage probability of product rejection from 0.5% to 0.25% was obtained.