基于变量选择的红酒质量分析

Analysis of Red Wine Quality Based on Variable Selection

摘要: 葡萄酒因其特殊的营养价值和较好的保健效果越来越受到消费者的欢迎，对葡萄酒质量的定量与定性分析是消费者关注的焦点。对Kaggle网站公开的红酒质量数据集开展了系统的研究，首先运用描述性统计分析方法，对数据集进行初步探索，直观呈现各变量的基本特征与分布规律；然后采用岭回归、LASSO、弹性网三种变量选择方法对该数据集建立多元线性模型，选出影响红酒质量的关键变量，该研究成果为红酒质量的精准检测提供了科学参考。

Abstract: Due to its special nutritional value and good health-care effects, wine is increasingly welcomed by consumers. The quantitative and qualitative analysis of wine quality has become the focus of consumers’ attention. A systematic study was carried out on the red wine quality dataset publicly available on the Kaggle website. Firstly, descriptive statistical analysis methods were used to conduct a preliminary exploration of the dataset, intuitively presenting the basic characteristics and distribution laws of each variable. Then, three variable selection methods, namely ridge regression, LASSO (Least Absolute Shrinkage and Selection Operator), and elastic net, were adopted to establish multiple linear models for the dataset, so as to select the key variables affecting red wine quality. The research results provide a scientific reference for the precise detection of red wine quality.