Skip to main content
ARS Home » Pacific West Area » Albany, California » Western Regional Research Center » Healthy Processed Foods Research » Research » Publications at this Location » Publication #336748
Title: Predictive model for consumer preference of a dried, chip-style persimmon product

Author
item Woods, Rachelle
item LAFOND, SEAN - University Of California
item Smith, Jenny
item Milczarek, Rebecca
item Preece, John
item Breksa, Andrew

Submitted to: Annual Meeting of the Institute of Food Technologists
Publication Type: Abstract Only
Publication Acceptance Date: 3/18/2017
Publication Date: 6/27/2017
Citation: Woods, R., Lafond, S.I., Smith, J.L., Milczarek, R.R., Preece, J.E., Breksa Iii, A.P. 2017. Predictive model for consumer preference of a dried, chip-style persimmon product. Annual Meeting of the Institute of Food Technologists. Session P05 Poster Session.

Interpretive Summary:

Technical Abstract: The State of California is a major producer of Asian persimmons (Diospyros kaki), however, there is limited availability of persimmons outside of this region and the fruit’s short harvest season. A dried, chip-style product could increase the geographic area and timeframe in which persimmon growers may sell their crop thereby increasing profitability. Taste, texture, and flavor are probable drivers of liking of dried fruit products. To determine which sensory qualities are predictive of consumer preference of the product, a model was created and validated. In fall 2015, dried chip-style products were prepared from 39 persimmon varieties and evaluated by a trained panel. During the season, 23 varieties were evaluated by a consumer panel. Using partial least squares (PLS) regression, a model was constructed to predict consumer liking from the sensory data. In fall 2016, the remaining 16 varieties were evaluated by a consumer panel. In both cases products were ranked in order of preference by 90 consumers using a balanced incomplete block design. The model predictions were compared to the consumer test in order to validate the predictive model. Two variations of the model were tested. The first variation included 4 taste (3 taste, 1 aftertaste), 7 texture, 21 flavor, and 1 overall flavor intensity attributes. The other contained all the texture and taste attributes, but only the single overall flavor intensity attribute. The PLS model using all the collected sensory data moderately predicted consumer preference (R2 = 0.58; Spearman’s rho = 0.66). The PLS model using only the single overall flavor intensity attribute well predicted consumer preference (R2 = 0.74; Spearman’s rho = 0.82). These results suggest that taste and texture drive preference and individual flavor attributes are likely secondary factors. In the more predictive model, sweetness, overall flavor intensity, and moistness were attributes associated with consumer preference while astringency, crispness, skin toughness, hardness, fibrousness, bitterness, and sourness were negatively associated. Interestingly, the traditionally positive attribute, crispness, was a negative driver for this product, as consumers tended to prefer moister chips.