Application Note: A Prediction Model for Determining Wine Variety
6 December 2012

In this application note a prediction model for determining wine variety is developed using the Agilent LC/MS Q-TOF and Mass Profiler Professional. Using Agilent’s Mass Profiler Professional software, single MS data containing over 26000 entities from the analysis of three varieties of wine are statistically evaluated. Once filtered on differences, principle component analysis shows that the wine variety can be grouped by specific compounds found in the wine samples. With this knowledge, a model based on partial least squares differentiation is made and unknown wines can be classified. All this is done without knowing the identity of the marker compounds that can distinguish one grape from another.