Abstract: The teas are classified into two categories—teas with chestnut-like aroma and other ones without chestnut-like aroma, according to the organoleptic evaluation results carried out by experienced experts. The screening process was conducted on the common components of teas with chestnut-like aroma along with the Bayes stepwise screening process, and set the aromatic components after screening as the variables. Then the discrimination equation was established. Results of cross-validation and validation of testing samples demonstrated that the correct rates of discrimination equation, established by cis-jasmone and geranylacetone, reached 94.44% in both discrimination analysis and cross-validation of training dataset, and the 8 testing samples are judged correctly, the correct rates of discrimination was 100%, Thus, it can be concluded that the discrimination equation worked effectively and the results was satisfactory. In addition, it can be seen that the chestnut-like aroma could be distinguished effectively by the selected variables in the principal components figure of cis-jasmone and geranylacetone.

Key words: chestnut-like aroma, Bayes stepwise discrimination, cross-validation, sample-validation, chemical recognition