In the ever-evolving geography of health care, prognosticating and precluding chronic conditions like diabetes are paramount. Numerous factors can cause a person to get affected by diabetes, like excessive abnormal cholesterol levels, body weight, family history, bad food habits, physical inactivity, etc. Among the most prevalent signs of this illness is increased urination. This paper presents a machine learning-based approach for diabetes prediction, using a rich dataset comprising especially blood glucose levels along with vital biomarkers, clinical history, and demographic details. Various algorithms are used and trained with our collected dataset. Among these algorithms, it was observed that random forest produced accurate results.