机器学习模型在白葡萄酒质量评价中的应用

Application of Machine Learning Model in Quality Evaluation of White Wine

摘要: 传统的葡萄酒质量检测由专业品酒师进行鉴评完成，存在检测成本高、周期长、主观臆断等缺点。建立一套客观、有效的葡萄酒质量评价体系，实现对葡萄酒质量的快速、批量检验是必要且重要的。本文对基于Logistic回归分析和随机森林两种白葡萄酒质量检测方法进行比较研究，选取了4898个白葡萄酒样本，通过混淆矩阵与十折交叉验证后，得出随机森林模型在测试集精确度及训练集精确度均优于Logistic模型，测试集精确度均值为88.48454%，相比于Logistic回归模型提高了8.36个百分点。本文使用机器学习模型为白葡萄酒的评价体系提供了一种快捷、准确且科学合理的方法。

Abstract: Traditional wine quality testing is conducted by professional wine tasters, which has drawbacks such as high testing costs, long testing cycles, and subjective assumptions. It is necessary and important to establish an objective and effective wine quality evaluation system to achieve rapid and batch inspection of wine quality. This paper conducts a comparative study of two white wine quality detection methods based on Logistic regression analysis and random forest, selects 4898 white wine samples, and through the confusion matrix and ten fold cross validation, it is concluded that the precision of random forest model in the test set and training set is superior to the Logistic model, and the average precision of the test set is 88.48454%, which is 8.36 percentage points higher than the Logistic regression model. This article provides a fast, accurate, and scientifically reasonable method for evaluating white wine using machine learning models.