One of the chronic and deadliest diseases that cause an increase in the sugar level in blood is diabetes. In fact, diabetes can inflict many severe effects like burning extremities, kidney & heart failures, myopia, and blurred vision. Diabetes develops when the body cannot produce enough insulin to maintain the threshold or when blood sugar levels are above a certain level. In HCS (Healthcare Services), Machine learning has gained a signification position because of its improving ability of disease prediction in healthcare services. The tedious identifying technique needs the patient should consult a doctor and visit a diagnostic center. This critical problem was solved by the rise of machine learning techniques. Hence there is a requirement for unified framework designing & diabetes prediction implemented by machine learning is proposed here. The three basic classifications of machine learning algorithms are Naive Bayes, Decision Tree, and support vector machine (SVM) used in this proposed system for detecting diabetics at the early stage. From the sources of UCI (University of California) machine learning respiratory the experiments are performed on PIDD (Pima Indians Diabetes Database). The three algorithms’ performance is evaluated by different parameters such as F-Measure, Accuracy, Recall, and Precision measured through classified instances incorrectly or correctly. From the obtained results the Naïve Bayes exhibits the highest accuracy of 78% compared to the remaining algorithms.