The study goes through two phases: first, attribute identification and selection, handling missing values; second, diabetes prediction model development using simple models for baseline and combining them to create classifiers using ensemble methods. The simple classifiers serve as baseline models, and their performance was enhanced by using ensemble techniques such as bagging, stacking and voting. These techniques combine multiple models to create robust and generalized classifiers. Our baseline models include SVM, KNN, random forests and decision trees. The results demonstrate that ensemble methods outperformed our baseline models in terms of accuracy and precision, highlighting the importance of integrating multiple models to improve diabetes prediction outcomes. This study contributes to the understanding of how advanced modelling techniques can be useful for diabetes management, focusing mainly on diabetes found in women. The paper makes use of the publicly available Pima Indians Diabetes dataset.