Flavor quality models for consumer acceptability using partial least squares regression

Abstract

Consumer acceptability models were developed for tomato and lemon as a function of sensory descriptors. Then these descriptors were modeled as a function of volatile compounds. The model for “acceptable”+“tastes great” (ATG) for the red tomatoes predicted most of the variability (79%-95%) with three variables. The models for “acceptable” (ACC) and “tastes great” (TG) needed five. The model for ATG for the red stage was validated. cis-3-hexenal, important in tomato flavor, was present in the models for the breaker but not for the red stage. Consumers detected differences in lemonade after 24h of storage. During storage citral and camphene disappeared, pinene, -bisabolene, trans- -caryophyllene, transocimene and -terpinene decreased; and -terpinene, -terpinolene and terpineol increased. The model for consumers response vs. sensory descriptors explained more than 94% of the variability for ACC using five descriptors, in the model for TG, four were enough for more than 87% of the variability explained.