PLSR modelling of quality changes of lager and malt beer during storage

Abstract

The aim of this research was to create mathematical models for describing the changes in beer properties by using two chemometric methods applied on the experimental data. The models are intended to be useful and trustworthy for calculating four beer properties based on three easily measured ones. For that purpose, lager and malt beer were packaged in glass bottles, while lager beer was also packaged in polyethylene terephthalate (PET). Samples were placed at room temperature in the dark for 6months. Fifteen physical and chemical properties of the beer were measured before bottling, immediately after bottling and once per month for the next 6months. Standard MEBAK and Analytica-EBC methods of analysis were applied. During the 6month period, seven properties changed >1%. Two partial least squares regression methods [polynomial regression, partial least squares regression with polynomial regression (PLSR-PR) and response surface method (PLSR-RSM)] were used for modelling the relationships amongst multivariate measurements. Models with high statistical significance were determined and two PLSR methods were compared. Both chemometric methods were found to be suitable for modelling physical and chemical changes in the beers during their commercial shelf-life. The PLS-RSM method was found to be the more precise and confident method in describing property changes for lager and malt beer in glass bottles, while the polynomial regression model was found to be better for the lager beer packaged in PET. The R2 values determined for polynomial regression model were up to 0.939, while for the random surface method model the values were up to 1.000.