Comparison of the red wine quality prediction accuracy using 5 Machine Learning Model

Abstract

Wine quality is rather important for alcohol industry, especially for some unique product (e.g., Vinho Verde). Since the quality evaluation of wine is often part of the certification process and is helpful for setting prices, it is necessary to find a model with the highest model accuracy to predict the quality. The paper collected the red wine data from a certain database and changed it to a classification problem with quality as 0 to bad and 1 to good. This dataset is like a normal distribution with the quality of 5 and 6 having the highest ratio. In order to find the approach with the best performance, five machine learning approaches are adopted and compared in terms of various metrics and judgments. According to the analysis, the Support-vector machine (SVM) model predicts preforms the best, i.e., the model has the highest accuracy. This article is a guide for the wine market for the model selection and shed light on guiding further exploration regarding to wine quality prediction.