Predictive Modelling of Diabetes Using Ensemble Classifiers: Findings from a Large and Integrated Data Set

In this research paper, we examine which of a number of machine learning classifiers are able to predict diabetes in a dataset of 100,000 cases where features include gender, age, state, binary health indicators, smoking, body mass index (BMI), and glucose level, one of which was known to be diabetic. Extensive exploratory data analysis was undertaken to explore the distribution and the relationships between features of the data set and these predictions. (Prajapati, (2023, July).We then evaluated the model output using metrics like as the Accuracy, AUC-ROC, Precision-Recall curves. Support Vector Machine (SVM), Voting Classifier (VC), and Stacking Classifier (SC). The output of these models was then evaluated with the help of measures using metrics like accuracy, AUC-ROC, and Precision-Recall curves. (Hasan, 2020). The highest accuracy of 98.75% was achieved by the voting classifier (VC), Random Forest (RF) was second with 97.18%, and XGBoost was 3rd with 97.15%. The results of this research can be applied in diabetes prediction, with an important conclusion on how ensemble methods could improve the accuracy indicators for healthcare applications.