Research On Wine Quality Identification Based On Data Mining

China is not only the largest wine producing country in the world,but also the largest consumer market of alcohol consumption.With the continuous development and progress of the information era,the wine industry is also gradually entering the automatic information system.In order to further improve the market competitiveness of China’s wine products,in the identification of wine,the artificial identification of the wine taster can not meet the requirements of today’s wine production.We should develop a scientific and efficient wine identification technology.This paper uses data mining classification algorithm to study the quality of wine.The classification of wine grade is carried out by data analysis of several main components of wine,so as to make the quality identification of wine scientific and efficient.First of all,this paper mainly introduces four kinds of data mining classification algorithms,which are KNN classification algorithm,Logistic regression algorithm,BP neural network and support vector machine algorithm,in which the support vector machine algorithm is optimized.The SMOTE algorithm is introduced because the original data is not balanced and the local classification results are not ideal.Data balance is processed to effectively improve classification accuracy of small data samples.In order to optimize the classification effect of wine grade,a better classifier is obtained.In this paper,a further optimized SVM-Bagging ensemble classifier is obtained by combining the optimal SVM classifier with the Bagging integration algorithm,which makes the classification effect improved and the SVM classifier and Bagging are integrated.The combination of law and practical classification of wine data sample identification is also the innovation of this paper.From the classification of wine sample data,various classifiers classify the original sample data preliminarily,and balance the sample data to make the data samples of a few classes are low from the prediction precision or even zero to the most class prediction accuracy,and the ensemble classifier further optimizes the effectiveness of the classification.Finally,the classification effect of classifier is much better than that of original classifier.