Predicting Wine Score based on Physicochemical Properties

Abstract

Existing wine scoring systems are based on subjective feelings, which are difficult to quantify. Thus, the score of a particular wine has little guidance on the brewing process. In this paper, regression tree and random forest are used to predict wine score by using the physicochemical properties of wine. According to the results of the model, higher quality wine can be produced by controlling the physical and chemical properties, thus reducing the waste of raw materials.