Application Of Multiple Machine Learning Algorithms In Diagnostic Classification Of Diabetes

In recent years,the rise of machine learning,deep learning,and artificial intelligence has brought about tremendous changes in people’s lives.At present,artificial intelligence has been applied to many areas such as intelligent control,image processing,and lan-guage recognition.In the field of medicine,use is still very rare.This article studies the prediction of diabetes using machine learning algorithms to find the optimal model for diabetes detection so that it can be applied to the medical field.This article uses the Pima Indian Diabetic Data Set in UCI,in the data after the preprocessing.Before the model is modeled,the cross-validation method is used to find the optimal parameters of each model.And the accuracy of the training set is used to find.After modeling,each model is calculated through the confusion matrix diagram.The F-score is used as the evaluation criteria of the models.The models chosen in this paper are single model SVM,KNN,integrated model Bagging,random forest,GBDT,and Voting for classification vot-ing model.Through comparison,find the most suitable forecasting model for data set Voting,followed by KNN,SVM and Bagging.At the same time,we found that for the prediction of diabetes,the importance of attributes degrees from high to low are:blood glucose level,age,BMI,genetic index,pregnancy frequency,blood pressure,insulin con-tent,sebum thickness.This study not only can complete the classification and detection of diabetes,but also plays a very important role in the control and prevention of diabetes.