Diabetes is an incessant illness and a significant general wellbeing challenge worldwide and adds to nerve harm, visual deficiency, coronary illness, expands the dangers of creating kidney sickness and coronary illness and vein harm. The fundamental goal of this work is to plan a classification model by utilizing the machine learning methods. Counts are done to anticipate diabetes in patients at a beginning phase with most extreme exactness by utilizing machine learning classification algorithm specifically SVM, Naive Bayes, Decision tree, Random Forest, Linear Regression, and K-NN, Neural Network. Dataset is taken from UCI (Machine Learning Repository) and calculations and tests are done on the dataset and result got shows Neural Net, improved k-NN, and improved Random Forest beats with most elevated precision of (96%) and (93%) and (78.8%) nearly different calculations.