Diabetes Prediction Using Different Ensemble Learning Classifiers in Machine Learning

Abstract

Abstract: Considered a chronic illness, Diabetes results due to increased level of the glucose in blood in the body which happens due to either less insulin production or if the response to insulin by body cells is not proper. Current practice in hospitals is to collect the required information for diabetes diagnosis through various tests and treatment is given based on the test results. Producing accurate results through prediction models of diabetes is difficult because there is not much data available and there is presence of outliers as well. This Literature proposes an optimal prediction model for diabetes where the raw data collected will go through few pre-processing techniques before introducing to the ML Classifiers such as Random Forest, AdaBoost, XGBoost. Using the pre-processing techniques and ensemble methods we have got better performance results. The weights of ML models are reviewed using their respective Area Under ROC Curve (AUC) result.