A REAL TIME CLOUD BASED MACHINE LEARNING SYSTEM WITH BIG DATA ANALYTICS FOR DIABETES DETECTION AND CLASSIFICATION

Shashikant, Anita S Harsoor

Abstract: Class one diabetes is one of the major health related issue observed across the globe almost 20 percent of worlds population is suffered by type2 Diabetes. Diabetes mellitus is a condition where body stops producing insulin which is responsible for Conversion of sugar into protein. Diabetes does not have any procure and is extremely difficult to identify at primary stage. Not only does it depends upon Sugar Content of body but also various other factors such as age, gender, geography, usage of particular medicine, so it is extremely difficult to diagnose. Therefore sugar content in the body keeps increasing which is harmful for various organs. In this work we present a novel category using Nave Bayesian classifier and KNN Classifier to classify a given set of diabetic parameters which includes Sugar observations such as Post Food, Fasting, and Average, and Gender, age of the patients record into Normal or Diabetic. Results shows that the system can classify diabetic status with an accuracy of Bayesian classifier 65.25% and 62.5 %of KNN classifier. Our system run from the cloud as an independent kernel and provides an opportunity to classify a given data from anywhere in the World.

Keywords: Bayesian classifier, KNN classifier, big data.

DOI: https://doi.org/10.15623/ijret.2017.0605020