The methodology for finding relationship between sensory data and instrumental data on beef textural properties


In the past, we usually employ the regression model for finding the relationship between sensory data and instrumental data. Here we would like to introduce a better method by employing PLS multivariate model.

The study was on comparing the textural properties of beef by 4 different thawing methods. The textural properties were collected by a texturmeter (instrument data) and a trained panel (sensory data with 14 panelists). The sensory test was conducted via magnitude estimation test which let the panelist to scale the sensory intensity with ratio scale. The data created by magnitude estimation test will be more similar to the instrumental data. There were 4 sensory attributes (hardness, juiciness, fibrousness, chewiness) and 7 instrumental indexes.

The results showed that coefficients for traditional regression models ranged from 0.5~0.6 while coefficients for PLS models ranged from 0.8~0.9. This expressed that PLS model can be better to explain the relationship between instrumental data and sensory data. This study also showed that well trained panel can produced stable and measurable data.