Modeling the Organoleptic Properties of Matured Wine Distillates

摘要

We present how the supervised machine learning techniques can be used to predict quality characteristics in an important chemical engineering application: the wine distillate maturation process. A number of experiments have been conducted with six regression-based algorithms, where the M5' algorithm was proved to be the most appropriate for predicting the organoleptic properties of the matured wine distillates. The rules that are exported by the algorithm are as accurate as human expert's decisions.