Brewed tea is a complex mixture of thousands of organic compounds, many of which change in response to environmental growing conditions and contribute to aroma, taste, and health beneficial qualities. Analyzing such a complex mixture by GC/MS is insufficient to identify low concentration compounds. Spectral deconvolution is a method of identifying many more compounds from GC/MS data that relies on a database built using clean spectra from GC-GC/MS data. Furthermore, MS subtraction can be used to identify even more co-eluting non-target compounds. This process often generates data with many more variables than samples, which presents a problem for many traditional multivariate analysis methods. A combination of PCA, an unsupervised method, and OPLS, a supervised method, allow for statistical analysis and identification of important metabolites using results generated by spectral deconvolution of GC/MS data.