DRAP: Decision Tree and Random Forest Based Classification Model to Predict Diabetes

This paper proposes a method called "DRAP" for detecting Diabetes. DRAP uses recorded data of the existing patients and classifies the unknown instances into one of the two classes - yes or no (whether diabetic or not). We have used a hybrid of decision tree and random forest algorithms to construct DRAP.Features considered for the constructing DRAP are glucose level, blood pressure, insulin, body mass index, age, etc. The proposed algorithm is evaluated on real-life data set. Our seminal results show the accuracy of 72% for Decision Tree and 76.5% Random Forest, respectively.