ABSTRACT
Large fraction of people is affected with diabetes all across the world and early detection plays an instrumental role to improve the survival chances. We propose a preliminary preventive measure, which can be introduced just by performing some clinical laboratory tests, results of which when fed into a machine learning classifier (XGBoost) can predict the possibility of diabetes in the concerned individual with an accuracy of almost 87%. In this paper we have worked with Logistic Regression, K Nearest Neighbors, Classification Tree, Random Forest Classifier, XGBoost Classifier, Adaboost Classifier and LGBM Classifier as the different Analysis of Machine Learning Algorithms for Prediction of Diabetes machine learning algorithms to identify which one among these would have the best accuracy and fine-tuned it with 10 folds cross validation to be used as an initial screening process to identify possible individuals having diabetes.
