Early detection and treatment of diabetes play an important role to keep people diagnosed with diabetes healthy. In this modern era of technology, machine learning can help us to detect diseases accurately. Classification is an important aspect of machine learning which is used for prediction. And this thesis aimed to explore some binary classification models, the finalized model of which can be used to predict diabetes in the early stage. To reach this aim, we were provided with a survey dataset containing reports of diabetes-related symptoms of 520 persons. Using survey data, we assessed the capabilities of machine learning models to predict patients at risk and evaluate highly associated symptoms among patients in the data that is correlated to diabetes. The trained model uses the classifiers to learns the symptoms of the patients and tries to predict whether the patient is likely to have diabetes. In the end, performance and quality measures of the classifiers evaluated using various performance metrics, and evaluations of the classifiers were compared.