Abstract: Sausage preference was predicted using self-organizing maps (SOM) based on sensory evaluation. The texture parameters and color data of 99 sausage samples were collected and correlated versus the sensory evaluation results using linear regression analysis. Principal component analysis (PCA) was used to eliminate the redundant data. An SOM model with the competition layer of 6 neurons and the output layer of 36 neurons was established. The results showed that the accuracy rate was 100% by extracting and classifying the eigenvalues of sausage samples. At this time, the root mean square error (RMSE) of the prediction set was 0.118 4, which implies that the model showed good generalization ability. This study aims to establish an accurate and efficient method for predicting food preference, which will provide useful data for new food product development and market preference prediction.

Key words: self-organizing maps, sausage, preference, prediction