Abstract: With the rapid development of information technology,the classical algorithm in computer has been extensively studied and applied in wine industry.The characteristics of machine learning algorithm are to use the technology of artificial intelligence,and a large number of samples in the set of training and learning can atuomatically identify the model and parameters that operation needs.Related research was used in data mining machine learning algorithms in this paper.The research of data mining technology based on classification algorithm was taken as an example.And for the weak generalization ability of SVM(Support Vector Machine),we proposed an improved SVM-NSVM,in which the training set is selected precisely according to each sample,similarities and differences between the subject nearest class choice is decided,and then SVM is trained to get classifier.For big disadvantage of the training data set in kNN’s(K-Nearest Neighbor),an improved progressive idea was given to find the nearest neighbor.Experiments show that,NSVM has more advantages than SVM in classification accuracy,speed classification.Complexity of the improved kNN algorithm is significantly reduced.In addition,the wine information and data analysis system was designed,and the data mining method was used to analyze,contrast and match the extremely large amount of wine information data so as to excavate the comparative information of the main components of wine and marketing potential information.Then these components was designed accordingly.With high-quality wine in the corresponding comparison of the ingredients,the final analysis analyze wine-related information and data can help wine producers analyze wine content and wine quality.

Key words: Machine learning,Data mining,Classification algorithm,Wine,Data analysis