Risk Prediction of Diabetes Based on Spark and Random Forest Algorithm

Abstract

Diabetes is one of the fastest growing global health emergencies in the 21st century. Therefore, this paper uses random forest model to improve the detection and prediction of diabetes. The data set containing age, gender, polydipsia and other attributes of 520 diabetic patients downloaded from the website of AIMCO is selected and transformed into a training set and a test set according to the ratio of 7:3. Based on the spark big data computing platform, the random forest model was constructed by the training set and the model was validated by the test set, and the model prediction performance was evaluated by observing the goodness of several indicators. The accuracy of the model was 0.9712, the precision was 0.9841, the recall was 0.9688, the AUC was 0.9719, and the F1-Score was 0.9714. Comparing the prediction evaluation indexes with other algorithms, it was found that the random forest algorithm was better than other algorithms, and the effectiveness of the model was verified. The important predictors identified by the random forest classification algorithm in the experiment can provide useful information for predicting the risk of diabetes mellitus to promote the development of disease prediction technology.