Wine quality grade data analysis and prediction based on multiple machine learning algorithms

With the continuous development of the wine industry, the study of wine grades has received more and more attention. Wine grade is one of the important indicators for assessing the quality of wine, which can reflect the characteristics of wine in terms of taste, aroma, colour and lustre, and can also guide consumers to make purchases. Therefore, the study of predicting wine grades is of great practical significance. In this paper, we first preprocess the data set, find out the outliers through box-and-line diagram and use the mean value to replace the outliers; secondly, we use Pearson correlation analysis to explore the correlation between the indicators of wine and the quality grade of grapes; lastly, by analysing the evaluation indexes of wine, we establish three machine learning models to predict the quality grade of wine. The experimental results show that the prediction accuracy of the three machine learning models for wine quality grade reaches about 60%, among which the random forest model has the best prediction effect, reaching 66.8%; the XGBoost model also has a better prediction effect, with an accuracy of 60.1%, and the decision tree model has a worse prediction effect, with an accuracy of 59.5%. By plotting the confusion matrix of predicted-actual values, it can be seen that the model performs better in predicting high grade wines but worse in predicting low grade wines. This study has certain reference value for the development of wine industry and consumers' purchase.