Diabetes Prediction Based on Random Forest Algorithm

Abstract

With the development of artificial intelligence, society will definitely become smarter in the future.More and more intelligent and convenient products will appear in everyone's life.For the medical industry, the huge amount of data generated is a valuable asset.How to discover the patterns of diseases in these data and improve the efficiency and accuracy of disease prediction through scientific means has become a hot issue for research nowadays.The main objective of this paper is to investigate the prediction of diabetes based on the RFA.In this paper, the RF algorithm (RFA) is used to analyse diabetes data, and the method is proposed based on the study of different DP models to improve the effectiveness of diabetes prediction (DP).Based on the RF-based ranking of feature importance, the features are added to the set of features to be evaluated in order of importance; for the proposed model is unable to perform effective feature selection, greedy feature selection based on RF is incorporated to further improve the accuracy of the DP model.The system was tested and experimented with, and the results showed that the prototype system of the DP model can effectively achieve the prediction function of diabetes.