The rising global prevalence of diabetes has prompted the medical field to explore innovative ways to enhance their diagnostic technologies. The primary focus of research in machine learning (ML) and deep learning (DL) is directed towards crafting sophisticated and effective systems for detecting diabetes. This research investigates the influence of cutting-edge machine-learning techniques and ensemble models on the forecasting of diabetes.Notably, accessibility to diabetes data is constrained, as the existing databases primarily consist of measurements obtained through laboratory-based and invasive testing methods [1].It is crucial to investigate anthropometric measurements and non-invasive tests to create a cost-efficient and highly effective solution[1].Numerous research studies suggest the possibility of creating detection models using anthropometric measurements and non-invasive medical indicators.In this paper, we have explored different machine-learning methods applicable to predicting diabetes detection. We have used five ML models - logisticregression, support vector machine (SVM), decision tree, random forest,and K nearest neighbor (KNN)on the diabetes dataset. We have also runsix ensemble machine learning models using averaging, maximum voting, stacking, blending, bagging, and boosting methods to evaluate the models' performance on the dataset.The performance of the models has been evaluated using RoC and precision-recall curves.