Diabetes Prediction Using Machine Learning Analytics
S. Reshmi, Saroj Kr. Biswas, Arpita Nath Boruah, Dalton Meitei Thounaojam, Biswajit Purkayastha
Abstract
Diabetes, an incurable disease which occurs because of high blood sugar levels over a prolonged time period, requires early prediction to significantly reduce its severity. Now-a-days Machine Learning (ML) community has been working on diabetes prediction and many researches have been done since decades for its prediction. Keeping in view the severity of this disease, this paper introduces a model, named Diabetes Expert System using Machine Learning Analytics (DESMLA), exploring the diabetes data to predict the disease more effectively. The diabetes dataset is imbalance in nature. And therefore, DESMLA model uses 5 most prominent oversampling techniques namely SMOTE, Borderline SMOTE, ADASYN, KMeans SMOTE, Gaussian SMOTE to get rid from this class imbalance problem of diabetes dataset. DESMLA model uses Decision Tree (DT) and Random Forest (RF) as classifiers along with all the data preprocessing steps for diabetes prediction. The experimentation results shows that DESMLA model with KMeans SMOTE and Gaussian SMOTE performs better.