Diabetes is normal and chronic condition. Age, obesity, sedentary lifestyle, genetics, poor diet, and hypertension are some variables that can cause it. Long-term diabetic patients are more prone to complications such as kidney disease, stroke, heart disease, eye disease, and nerve damage. The proposed course for diabetes is determined by diagnosis, which is usually made in hospital through medical testing. Health activities undergo a change with artificial intelligence, machine learning and integration of data analysis. Scientists continue to learn how to cope with their problems more accurately and correctly. Patients can take corrective measures in time when diabetes is identified and diagnosed quickly. Grade Bossing Classifier (GBC), Random Forest (RF), Support vector Machine (SVM), K-Nearest Neighbor (KN), Decision Tree, and Logistic Regulation (LR) Machine are among the learning models, which we protest in the thesis of this study. The eight most important variables—the blade print, insulin, skin thickness, diabetes, age, BMI, glucose and pregnancy—contrast with models. A clinical and demographic data set was used to assess the performance of the model, and KNN was found to be the highest accuracy of 82%. Conclusions show the importance of choosing the model suitable for predicting diabetes and KNN’s appropriate. The hybrid approach can be emphasized in the latter work, and functional engineers can improve the prognosis.