Selecting crackling product based on sensory analysis by different statistical data approaches Documento de conferencia uri icon

Abstracto

  • Cracklings, which is a well-known product in Mexico, are obtained by frying the pork skin. Up to know no attempt to formulate chicken cracklings has been done. Therefore, in order to take advantage of chicken by- products, during this experiment two different chicken cracklings prototypes were developed and compared with the pork ones. When a food prototype is ready, sensory analysis, which is related on how a food product is appreciated by the human senses must be performed. Which makes the consumers' acceptance a key for achieving an economical success in the food industry. In this paper four different products are analyzed: pork or chicken crackling, with or without sauce. For this analysis the acceptance of these products was tested by each consumer based on their perception (hedonic scales) with values ranging 1 to 10. In order to understand the distribution of the consumers' grading, a dimensionality reduction technique based on evolutive algorithms that plot the consumers' in a 2D-plane based on their grades distances was proposed and compared with PCA. To reinforce this understanding, the distance matrix and the dendogram of hierarchical clustering were used. A Liking Product Landscape is proposed, where the distribution of the product grades and of the consumers are shown in the same graph. The most accepted products are the ones with sauce, in particular the pork crackling product was the most accepted one.

fecha de publicación

  • 2017

Palabras clave

  • Backpropagation Algorithm
  • Classification Problem
  • Crossover Operator
  • Diverse Populations
  • Dot Product
  • Error Vector
  • Evolutionary Algorithms
  • Phenotypic Space
  • Regression Problem
  • Semantic Vectors
  • Sigmoid
  • Supervised Learning Problem
  • Tournament Selection
  • Unit Vector
  • Bayesian Classifier
  • Fitness Function
  • Gene Regulatory Networks
  • High-dimensional
  • Scalar Value
  • Selection Techniques