Relationship Between Sensory Characteristics of Cigarettes and GC/MS Data of Cut Filler Extracts

In order to predict the sensory characteristics of cigarettes with the GC/MS data of cut filler extracts, the samples of 48 commercial cigarette brands were panel tested and their cut fillers were extracted with accelerated solvent extraction, and the resultant extracts were analyzed by GC/MS. The quantitative models were developed by processing the GC/MS data and panel test data with principal component regression (PCR) and partial least square (PLS). The results showed that the PCR models offered better predictive effects than did the PLS models. The mean absolute errors between the predictive results of PCR models and the values of panel test were below 0.5 for more than 75% of sensory characteristics. It was concluded that PCR models were reliable and could be used as a method for predicting the sensory characteristics of cigarettes.